Next Article in Journal
Re-Examination of the Relationship between Industrial Agglomeration and Haze Pollution: From the Perspective of the Spatial Moderating Effect of Environmental Regulation
Previous Article in Journal
Opportunities for Prediction Models to Reduce Food Loss and Waste in the Postharvest Chain of Horticultural Crops
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 16
Issue 17
10.3390/su16177805
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

A Review of Research on Building Energy Consumption Prediction Models Based on Artificial Neural Networks

by
Qing Yin
1,2,
Chunmiao Han
1,2,
Ailin Li
1,2,
Xiao Liu
1 and
Ying Liu
1,2,*
1
School of Architecture, Harbin Institute of Technology, Harbin 150001, China
2
Key Laboratory of Cold Region Urban and Rural Human Settlement Environment Science and Technology, Ministry of Industry and Information Technology, 66 Xi Dazhi Street, Harbin 150006, China
*
Author to whom correspondence should be addressed.
Sustainability 2024, 16(17), 7805; https://doi.org/10.3390/su16177805
Submission received: 15 August 2024 / Revised: 3 September 2024 / Accepted: 5 September 2024 / Published: 7 September 2024
Browse Figures
Versions Notes

Abstract

:
Building energy consumption prediction models are powerful tools for optimizing energy management. Among various methods, artificial neural networks (ANNs) have become increasingly popular. This paper reviews studies since 2015 on using ANNs to predict building energy use and demand, focusing on the characteristics of different ANN structures and their applications across building phases—design, operation, and retrofitting. It also provides guidance on selecting the most appropriate ANN structures for each phase. Finally, this paper explores future developments in ANN-based predictions, including improving data processing techniques for greater accuracy, refining parameterization to better capture building features, optimizing algorithms for faster computation, and integrating ANNs with other machine learning methods, such as ensemble learning and hybrid models, to enhance predictive performance.

1. Introduction

The International Energy Agency’s (IEA) Global Buildings Status Report highlights the continuous increase in energy consumption, with building operations accounting for 30% of global final energy consumption in 2021. To address this, building energy consumption predictions can inform energy-saving measures and strategies, ultimately leading to reduced energy use and carbon emissions.
Building energy consumption prediction is a crucial tool for assessing energy-saving potential throughout a building’s design, operation, and retrofitting phases. Common prediction methods include model-driven (white box), data-driven (black box), and hybrid (grey box) approaches. Data-driven methods, particularly machine learning, have garnered significant attention due to their time efficiency, ease of operation, and relatively accurate prediction performance. Artificial neural networks (ANNs), as a versatile machine learning technique, are widely considered one of the most effective methods for building energy consumption prediction [1]. Since 2016, researchers have delved deeper into this field, exploring various algorithm optimizations, model integrations, and hybrid approaches to enhance ANNs’ predictive power.
Several recent review articles focus on building energy consumption prediction using ANNs. Nadia D. Roman et al. [2] explored the creation of ANN-based meta-models for building performance simulation (BPS); Chujie Lu et al. [3] analyzed open issues and challenges in applying twelve ANN architectures; Dimitri Guyot et al. [4] examined neural network applications, technical features, and limitations in architecture; Siti Solehah Md Ramli et al. [5] compared ANNs to other data-driven models using evaluation metrics; Saeed Reza Mohandes et al. [6] observed rapid-growth in ANNs for building energy analysis (BEA), particularly with GDBP and LMBP algorithms, and a shift towards newer neural network types (GRNN and RNN); Jason Runge et al. [7] noted the prevalence of black-box feedforward neural network models with manually determined parameters.
Through the reviewed literature, it is evident that building energy consumption prediction using artificial neural networks (ANNs) spans multiple disciplines, including architectural engineering, computer science and artificial intelligence, energy engineering, and data science. However, there is a notable gap in systematic reviews from an architectural perspective on the application of ANNs across different building phases. Consequently, this literature review aims to provide a thorough analysis of studies from 2015 to 2023, elucidating the diverse applications of various input types, building types, energy types, and temporal characteristics in energy consumption prediction models. It also proposes guidelines for selecting appropriate ANN structures for different building phases. Finally, this review discusses future potential developments in ANN-based predictions, focusing on enhancing data processing, refining model parameters, and integrating learning approaches. It emphasizes that improved data cleaning and processing can enhance data quality, optimized algorithms can refine model parameters, and the application of ensemble and hybrid models can improve model interpretability and predictive performance.
Figure 1 illustrates the continuous increase in publications on “Building Energy Consumption Prediction Using Artificial Neural Networks” from 2015 to 2023 (starting in 2017), indicating a significant research hotspot. This paper addresses this trend by reviewing the past decade of the literature on using ANNs to predict building energy use and demand. It summarizes how different ANN structures are applied in building design, operation, and retrofitting. Additionally, it discusses input types, building types, energy consumption types, and temporal characteristics of prediction models, ultimately offering guidance on selecting appropriate ANN structures for different building phases. Finally, this paper explores potential future developments in ANN-based building energy consumption prediction, considering data processing, model parameterization, algorithm optimization, integrated learning, and hybrid models.
This study employed bibliometrics, using the keywords “Neural networks”, “energy”, “prediction”, and “build” in search strings within the established and reputable Web of Science database. We selected English publications from 2015 to 2023. Initially, 2155 publications were found; after screening titles and abstracts, 549 were selected; and upon further review, 292 publications were finally chosen. The categorization of the literature is shown in Figure 2, with 38 review articles on building energy consumption prediction, of which only 5 are reviews on predictions using artificial neural networks, as depicted in the flowchart in Figure 3.

2. Analysis of Current Applications and Characteristics of ANNs

2.1. ANN

Artificial Neural neural Networks networks (ANNs) embody a category of machine learning algorithms inspired by biological neural networks. These models are noted for their robust ability to represent and model nonlinear relationships between inputs and outputs.
Initially developed in the late 1950s, artificial neural networks aimed to mimic neuronal behaviors in the human brain [8]. In 1995, Syed M. Islam [9] and colleagues leveraged feedforward neural networks to model and predict building thermal loads, laying a crucial groundwork for future research. By 2000, Kalogirou [10] highlighted the potential of ANNs in designing diverse energy systems. Subsequently, in 2005, González and Zamarreño [11] utilized feedback neural networks to forecast hourly electric energy use in office buildings. Following this, in 2006, Karatasou and Santamouris [12] promoted the use of feedforward neural networks (FFNNs) for building energy consumption forecasts. Subsequent research consistently validates that ANNs provide precise predictions of building energy use, with FFNNs noted for their high accuracy and widespread application in a multitude of studies.
With advancements in deep learning, Recurrent recurrent Neural neural Networks networks (RNNs) have emerged as potent tools for processing sequential data in building energy consumption prediction. In 2016, Marino, Amarasinghe [13], and Manic noted Long long Shortshort-Term term Memory’s memory’s (LSTM’s) challenges with minute-by-minute predictions but acknowledged its adequacy for hourly forecasts. The following year, Heng Shi [14] et al. introduced a pooling-based deep RNN, enhancing household load predictions. By 2018, Rahman [15] proposed using an RNN sequence-to-sequence model for mid- and long-term forecasting, which showed promising accuracy. Additionally, in the same year, Kumar, Hussain, Banarjee, and Reza [16] demonstrated that LSTM models, capable of addressing nonlinearities and retaining historical data, outperform traditional BP neural networks in predicting electrical grid loads.
In 2019, Kim and Cho [17] combined Convolutional convolutional Neural neural Networks networks (CNNs) with LSTM to predict residential power consumption, marking a significant application in energy modeling. The following year, Yuan Gao [18] and his team utilized a sequence-to-sequence model with a 2D CNN featuring an attention layer, enhancing prediction accuracy for specific buildings. In 2021, Guannan Li [19] employed a CNN-LSTM hybrid network, noting a significant reduction in computation time when integrated with attention mechanisms, though without accuracy improvements. By 2022, Ibrahim Aliyu [20] introduced a 1D CNN, celebrated for its computational efficiency, high performance, and cost-effective hardware demands. CNNs are relatively new to the field of building energy consumption prediction, yet they have garnered increasing attention with the continuous advancement in deep learning technologies.
Prior to 2017, publications on artificial neural networks were somewhat sparse. However, post-2017, there has been a substantial surge in the literature related to building energy consumption prediction using artificial neural networks, with as many as 474 studies published in 2022. In summary, ANNs are extensively applied in building energy consumption prediction, becoming increasingly valued by researchers and representing one of the most commonly used prediction algorithms today. ANNs are categorized into FFNNs, RNNs, and CNNs. As depicted in Figure 4, publications from 2015 to 2023 on ANNs, FFNNs, RNNs, and CNNs show a continuous increase. FFNNs are the most frequently utilized neural network structure. Publications on RNNs have gradually increased since 2018, while research on CNNs has only recently gained public exposure, indicating significant potential for further exploration.

2.2. Characteristics of Neural Networks by Structure

2.2.1. Feedforward Neural Networks (FFNNs)

Feedforward Neural neural Networks networks (FFNNs) represent a fundamental and widely used type of neural network architecture. Their structure consists of an input layer, one or more hidden layers, and an output layer. Key FFNN types include Multilayer multilayer Perceptron perceptron (MLP), Radial radial Basis basis Function function (RBF), Extreme extreme Learning learning Machine machine (ELM), Wavelet wavelet Neural neural Network network (WNN), and Nonlinear nonlinear Autoregressive autoregressive with Exogenous exogenous Input input (NARX). Table 1 summarizes their characteristics, advantages, disadvantages, and structures. Among them, MLPs are the most widely used and versatile, known for their ability to model and learn complex nonlinear relationships. MLPs perform well across various tasks, such as regression, classification, and pattern recognition. RBF networks are suited for regression and classification problems, while WNNs excel in signal processing and time series analysis. ELMs are designed for large-scale data and high-dimensional features, offering fast training times. NARX models, which take into account past observations and external inputs, are well suited for time series forecasting, particularly for data with periodic or trending characteristics. Each type of FFNN has its own structure, characteristics, and strengths, and selecting the appropriate model depends on the specific problem, data characteristics, and desired outcomes.

2.2.2. Recurrent Neural Networks (RNNs)

A Recurrent recurrent Neural neural Network network (RNN) features a unique architecture with recurrent connections tailored for processing sequential or temporally dependent data. The foundational RNN structure comprises an input layer, a hidden layer, and an output layer, where outputs from the hidden layer are recurrently fed back as inputs, creating a loop that facilitates data flow across time steps. Key variants of RNNs, such as LSTM, Gated gated Recurrent recurrent Unit unit (GRU), Elman Neural neural Network network (Elman), Restricted restricted Boltzmann Machine machine (RBM), and Evolutionary evolutionary Data data Analysis analysis (EDA), are detailed in Table 2, highlighting their features, pros, cons, and architectural nuances. LSTM, a flexible and robust RNN variant, excels in managing tasks that involve long sequence data and require the capture of extended temporal dependencies. The GRU model, a streamlined version of LSTM, is optimized for rapid training and high computational efficiency, making it ideal for time-sensitive tasks. The Elman network, one of the simplest RNN forms, is well suited for basic sequence modeling tasks. Recurrent neural networks, in general, are adept at handling both sequential and recursive data structures. Bidirectional RNNs, in particular, are effective for tasks that necessitate an understanding of contextual relationships within data sequences, offering a deeper comprehension of bidirectional dependencies. Selecting an appropriate model involves a thorough evaluation of the specific task demands, data traits, and algorithmic efficacy.

2.2.3. Convolutional Neural Networks (CNNs)

Convolutional Neural neural Networks networks (CNNs) are advanced deep learning models, excelling in processing data with grid-like structures, including images and audio. These networks utilize a combination of convolutional layers, pooling layers, and fully connected layers to progressively distill and amalgamate features from the input data, culminating in performing classification or regression tasks via the fully connected layers. Given that building energy consumption data typically manifest as time series, CNNs adeptly discern long-term dependencies and periodic variations within such data. Through their convolution and pooling operations, CNNs are capable of extracting features across various time scales, thereby facilitating a more nuanced understanding and prediction of building energy consumption trends. Moreover, the prediction of building energy consumption often requires integrating multiple types of data, such as weather conditions and building attributes. CNNs address this by processing diverse data modalities concurrently through a multi-channel input layer, which enhances the fusion and learning of multimodal data features, thereby boosting the accuracy of predictions. Additionally, the spatial structure and layout of buildings play a critical role in energy consumption prediction. CNNs effectively map local features and spatial correlations within architectural structures using convolution operations, which enhances understanding of the spatial dynamics influencing building energy consumption and improves the accuracy of predictions.
Convolutional Neural neural Networks networks (CNNs) hold broad prospects for application in building energy consumption prediction. They can accurately predict and optimally manage building energy by processing time series data, integrating multimodal data, and handling spatial information.

3. Applications of Building Energy Consumption Prediction Models at Different Stages

“The entire lifecycle of a building” encompasses the complete process from design and construction to operation, maintenance, renovation, and even demolition. Within this lifecycle, energy consumption is a critical consideration, as the design of and operational approaches to a building directly impact energy efficiency and consumption. Energy consumption prediction throughout the building’s lifecycle requires an understanding of the characteristics and influences at different stages—for example, architectural structure and material choices during the design phase, and user behavior and equipment efficiency during the operational phase. Although there is existing research on building energy consumption prediction, comprehensive studies across different phases of a building’s lifecycle are relatively scarce.
Neural network models for predicting building energy consumption are predominantly applied across the design, operation, and renovation phases of a building’s lifecycle. Analysis of the initially reviewed literature shows that 11% of studies focus on the early design phase, 8% on energy-saving optimization design, and a significant 81% on the operational phase (Figure 5).

3.1. Early Design Stages

3.1.1. Energy Forecasting in Early Design Phases

The ANN training datasets primarily consist of input and output parameters, with inputs including building form, envelope structure, environmental conditions, human activity, historical energy usage, equipment efficiency, and date/time; outputs comprise cooling loads, heating loads, lighting energy, total energy, and electricity usage. These datasets predominantly derive from real data, simulated data, and benchmark data. The lLiterature review indicates that during the building design phase, research focusing on ANN-based energy consumption prediction models is primarily concentrated on residential buildings. Buildings are categorized by their functionality related to user activities, with residential buildings representing 50% of the research focus, followed by 14% on office buildings, 4% on rural buildings, 4% on campus buildings, 7% on commercial structures, and 21% on other types such as industrial and sports facilities. The dataset primarily comprises simulated data from parametric modeling post-energy simulation, accounting for 93%, with a small fraction derived from real historical data. Key building evaluation metrics include Mean mean Absolute absolute Percentage percentage Error error (MAPE), Root root Mean mean Square square Error error (RMSE), Coefficient coefficient of Variation variation of the Root root Mean mean Square square Error error (CV-RMSE), Mean mean Absolute absolute Error error (MAE), Mean mean Bias bias Error error (MBE), Mean mean Squared squared Error error (MSE), Coefficient coefficient of Determination determination (R2), and Mean mean Relative relative Error error (MRE), with MSE being the most extensively used, accounting for 30%, and MBE and MRE the least utilized, each at only 1%, as depicted in Figure 6. According to Table 3, input parameters are largely focused on enclosure structure and architectural form, while output parameters are mainly on cooling and heating loads.
The primary purpose of energy consumption prediction in the early stages of architectural design is to design the building’s form, identify the main factors affecting energy consumption, choose appropriate enclosure structures, and budget for energy costs, necessitating the collection of extensive datasets. Real data from completed buildings are difficult to collect and lack continuity, making the datasets incomplete and inadequate for training artificial neural networks. Therefore, in the early stages of architectural design for energy consumption prediction, it is common to first select input parameters with strong relevance for parametric modeling, then acquire corresponding output parameters through energy consumption simulations, and after processing the data, input the dataset into an ANN model for training and validation to develop an ANN-based building energy consumption prediction model. Subsequently, optimizations are considered along with costs to obtain the optimal solution, further determining parameters such as architectural form and enclosure structures to achieve the building’s optimal energy-efficient form and enclosure combination.
Utilizing Artificial artificial Neural neural Networks networks (ANNs) to predict building energy consumption in the initial phases of architectural design can lead to the development of an energy-efficient building prototype. This approach not only optimizes energy-saving measures during the conceptual design phase but also enhances energy efficiency, manages overall building costs, and ensures economic viability for long-term operations and maintenance.

3.1.2. Selection of ANNs in Architectural Design Phases

According to Figure 7, feedforward neural networks (FFNNs) dominate the literature on predicting building energy consumption during the design phase, with the introduction of Convolutional convolutional Neural neural Networks networks (CNNs) gaining traction since 2019. In 2020, X.J. Luo [47] and colleagues proposed three multi-objective prediction frameworks utilizing machine learning technologies—ANNs, support vector regression, and long short-term memory networks—to predict multiple types of energy consumption concurrently. Comparative analyses indicate that ANN-based models excel in both accuracy and processing efficiency.
For energy consumption prediction in the architectural design phase, the FFNN structure is commonly employed due to its suitability for non-temporal data, such as enclosure structure parameters and architectural form parameters, which lack time dependencies, making RNNs inappropriate for this phase. FFNNs are typically chosen, with CNNs selected on rare occasions. CNN models, with their capability for feature extraction and multi-scale prediction, are effective not only in extracting features of a building’s appearance, structure, and layout, but also in analyzing these features at different levels—including structural components, material selection, and spatial layout—by combining them for comprehensive analysis. The feature images of the buildings are then fed into the CNN model to predict energy consumption. Future research might explore using CNNs for building energy consumption prediction.

3.2. Operational Phase of Construction

3.2.1. Application of Energy Forecasting in the Operational Phase

The bulk of current research on “Building Energy Consumption Prediction Based on Artificial Neural Networks” is concentrated on the operational phase, representing 81% of the studies. In this phase, 26% of research targets residential buildings, followed by office buildings at 23%, public buildings at 5%, campus buildings at 16%, commercial buildings at 12%, and other building types such as industrial and sports facilities at 18%. Most studies focus on residential and office buildings, with significant attention given to variables like historical energy consumption, weather conditions, and timing. Given that most subjects of these studies are intelligent buildings, collecting real-time operational data—including temperature, humidity, air quality, lighting intensity, and occupancy rates—is straightforward. This data collection primarily comprises actual data, accounting for 85% of the datasets utilized. However, challenges remain with data gaps, discontinuities, and abrupt changes, directing the majority of research towards short-term predictions, which represent about 70% of all studies. Predictions typically center on electricity usage and overall energy consumption. Commonly used evaluation metrics include MAPE, RMSE, CV-RMSE, MAE, MBE, MSE, R2, and MRE, with MAPE being the most prevalent at 26% usage and MRE the least at only 1%, as depicted in Figure 8.
“Building Energy Consumption Prediction Based on Artificial Neural Networks” research is most extensive during the operational phase as buildings are active and generating substantial data. This phase provides a robust dataset, including historical energy usage, weather conditions, and equipment operations, forming a solid foundation for model development. Researchers often employ real-time data in neural networks, enhancing the credibility and accuracy of their predictive models. Predicting energy consumption during this phase helps facility managers monitor energy use efficiently, swiftly detect deviations, and implement optimization strategies. This not only curtails energy usage but also boosts efficiency, reduces operational costs, and elevates the economic performance of buildings.
An analysis of how the literature on the operational phase of building energy consumption prediction from 2015 to 2023 is distributed reveals a predominance of feedforward neural network-based studies at 43%, with recurrent neural networks following at 29%, and convolutional neural networks at 4%. Feedforward neural networks such as MLP, WNN, ELM, NARX, and RBF are prevalent, with MLP being the most common model used for energy prediction. Recurrent neural network studies are often centered around LSTM, GRU, and Elman networks, with LSTM featured most extensively. Convolutional neural networks are less frequently used compared to other types, but there is a growing trend towards hybrid models and ensemble learning, reflecting a broader diversification in research approaches (Table 4).

3.2.2. Selection of ANNs in the Operational Phase

In recent studies, Lei Gao [134] et al. compared MLP, LSTM, and CNN neural networks, finding that LSTM exhibited the best predictive performance, while MLP was the least effective. Raghavendra Chalapathy et al. [135] contrasted shallow learning models with deep learning models, discovering superior short- and long-term predictive capabilities in LSTM-based RNN-MIMO models. Lei Xu et al. [144] introduced three Bayesian deep neural network models—Recurrent recurrent Neural neural Networknetwork, Long long Shortshort-Term term Memorymemory, and Gated gated Recurrent recurrent Unitunit—among which the Bayesian LSTM (BLSTM) model achieved optimal performance. Yun Duan et al. [186] presented the LSTM-PQR model, which outperformed BPNN in predicting energy consumption in office buildings. Continuous research validates that LSTM surpasses feedforward neural networks in predictive accuracy due to its ability to effectively manage sequential data and capture long-term dependencies, typical of building energy data. Moreover, LSTM’s memory cells can retain and retrieve past information as needed, effectively addressing gradient vanishing and explosion issues, thereby enhancing its capability to model long-term dependencies. In contrast, feedforward neural networks struggle with long-term dependencies. Additionally, building energy consumption prediction involves multiple influencing factors such as weather, building characteristics, and equipment operational status, constituting multivariate time series data. LSTM proficiently handles such data, better delineating the interrelationships among various factors, thereby improving prediction accuracy.
According to Figure 9, recent years have seen a steady increase in research articles on RNN-based building energy consumption predictions, surpassing those based on FFNN. Concurrently, the literature on CNN and hybrid models for predicting building energy consumption continues to grow. Jingyi Zhou et al. [168] believe CNNs can effectively predict energy consumption throughout a building’s entire lifecycle with greater accuracy. Guannan Li et al. [19] conducted comparative assessments of LSTM and its hybrids, such as LSTM-CNN and CNN-LSTM, finding that CNN-LSTM models halve computation times and enhance prediction accuracy. Pingping Chen et al. [200] introduced an attention-based CNN-LSTM model that effectively retains historical information, thereby improving predictive performance for short-term load forecasting in smart buildings. Increasingly, researchers are validating the superior predictive capabilities of CNNs and related hybrid models. Consequently, during the operational phase of buildings, it is advantageous to prioritize RNN-based models for energy consumption predictions when datasets exhibit strong time series characteristics, with potential future exploration towards CNNs and hybrid models.

3.3. Renovation Phase

3.3.1. Application of Energy Consumption Prediction in Renovation Phases

In the building retrofitting phase, research focuses primarily on residential buildings (55%) and office buildings (25%). Smaller shares of research concentrate on rural buildings (5%), campus buildings (10%), and commercial buildings (5%). As Figure 10 illustrates, studies on energy consumption prediction models using artificial neural networks heavily target residential and office buildings. Design variables analyzed during the retrofit phase often center on enclosure structure and environmental parameters.
Data collection poses a challenge in this field. Most research on neural network models for building retrofit prediction relies on simulated data (89%), as collecting historical energy consumption, temperature, humidity, and wind speed data from existing buildings (without pre-installed sensors or flow meters) is difficult. Only 11% of studies utilize real-world datasets. Predicted types of building energy consumption typically include total energy consumption, and heating and cooling loads.
Common evaluation metrics include MAPE, RMSE, CV-RMSE, MAE, MBE, MSE, R2, and MRE. Of these, MSE is most widely used (31% of studies). During the building retrofit phase, energy consumption prediction is often paired with multi-objective optimization to achieve multiple goals following the retrofit. Optimization targets primarily focus on energy consumption, thermal comfort, cost, and lighting. As architectural form parameters are difficult to adjust during retrofits, variables typically concentrate on enclosure structures and equipment layout.

3.3.2. Choice of ANNs for Building Renovations

During the building retrofit phase, the application of artificial neural networks (ANNs) for predicting energy consumption facilitates the assessment of energy efficiency improvements and cost savings pre- and post-retrofit. This assessment provides a solid foundation for evaluating retrofit proposals, informing design choices, and identifying optimal retrofit strategies. Further, analysis of the prediction outcomes enables the fine-tuning of retrofit strategies, thereby maximizing both energy efficiency and economic returns.
In the retrofit phase, ANNs function as meta-models. The process begins with parametric modeling using initial datasets, followed by the acquisition of target parameter data via Building building Performance performance Simulation simulation (BPS) technology. Preprocessing steps such as data cleaning and normalization are then undertaken to ready the data for model training. Objective functions and constraints are subsequently established for specific multi-objective optimization tasks that tailor the model’s structure and parameters for optimized training. Once trained, the model forecasts future energy consumption, offering decision support and directing the implementation of energy management and retrofit measures. This workflow demands an ANN architecture that not only delivers precise predictions but also accommodates swift feedback and real-time adjustments.
As indicated in Table 5, FFNNs are the prevalent choice for predicting energy consumption in building retrofit scenarios, attributed to their straightforward and intuitive nature, which simplifies understanding and implementation. During the retrofit phase, the absence of complex time series data or significant long-term dependencies makes FFNNs ideal due to their rapid training capabilities compared to RNNs and other models. This allows for expedited training and predictions, crucial in retrofit contexts that demand quick feedback and immediate adjustments. Moreover, FFNNs’ adaptability makes them particularly effective in retrofit phases characterized by costly and lower-quality data collection that encompasses diverse data types and predictive tasks, streamlining the model selection and implementation process.
In summary, the selection of FFNNs for predicting energy consumption during the building retrofit phase is justified by their simplicity, swift training speed, wide applicability, and minimal data requirements. Nonetheless, the decision to employ a specific model type should be thoroughly assessed based on the actual conditions and specific demands of the prediction task at hand.

4. Future Research Directions

The process for predicting building energy consumption unfolds through a series of meticulous steps: initially, gathering data concerning the building’s architecture, weather conditions, historical energy usage, and equipment specifics. Once the data are processed, features are selected, and the input and output parameters are established. These data are then segmented into training, validation, and testing subsets. Subsequently, a neural network model is constructed and undergoes evaluation and validation after training, with adjustments made based on the validation outcomes. Ultimately, this trained model is implemented in real-world applications to forecast energy consumption in buildings. This entire procedure, depicted in Figure 11, is primarily influenced by data quality, feature selection, choice of model, and optimization strategies. Employing proper data handling, feature engineering, model selection, and optimization enhances the accuracy and dependability of the predictions.
Reflecting on the prevailing research trends, the future trajectory of this field can be segmented into three principal areas: firstly, enhancing the data preprocessing techniques by choosing methods apt for various stages; secondly, selecting optimal neural network architectures and fine-tuning model parameters during the model building phase; and thirdly, choosing effective algorithms for optimization. These measures are targeted to refine the overall predictive process and elevate the outcomes for building energy consumption predictions.

4.1. Data Processing

Data processing plays a pivotal role in predicting building energy consumption, critically affecting the models’ accuracy and validity. It forms an essential phase in the development of predictive models. Future research will benefit from advancing data cleaning, feature engineering, and integration techniques to boost data quality and model performance. Advanced strategies, including outlier detection and management of missing data, alongside feature selection and dimension reduction, are vital for enhancing both data integrity and predictive efficacy.
In the initial data cleaning phase, measures are implemented to identify and correct anomalies to ensure data accuracy and consistency. Established methods for outlier management include Chauvenet’s criterion [235], GRUBS outlier detection [236], and isolation forest (ISF) [190] for handling missing data and excluding outliers. To address incomplete preliminary data, methods such as transfer learning [60,128,129,143,153,171,175,237], exponential moving average (EMA) [142], XGboost [119], and decision tree classification (DTC) [119] are proposed. During the data cleaning phase, feature selection and extraction are necessary to reduce redundancy and lower dimensionality, thereby enhancing modeling efficiency and accuracy. Proposed methods include PCA [88,115,117], random forest (RF) [170,183], rough set theory [48,238], statistical moments [50], Spearman correlation coefficient [51,175,176], sensitivity analysis [68,88,169,172], regression analysis [58], stacked auto-encoder (SAE) [122], recursive feature elimination (RFE) [119,235], Kalman filter (KF) as a stochastic filtering method [208], differencing for effective information removal [237], Pearson correlation analysis [56,236,239], and mean impact value (MIV) for information filtering [103]. Additionally, data grouping or classification into sets with similar features involves using clustering analysis techniques to identify patterns within the data, thus better understanding the data’s structure and distribution. Techniques include clustering residuals based on Chebyshev distance [51], affinity propagation (AP) clustering [126], K-means, and K-nearest neighbors (KNN) methods [235].
The aforementioned studies have effectively addressed issues related to data incompleteness, selection, clustering, and outlier management, significantly improving the accuracy of building energy consumption predictions. In the operational phase of buildings, training data primarily consist of real-world data, with most studies relying on data collected from smart meters, while only a few use specialized metering equipment, which greatly impacts the accuracy of operational phase data. In contrast, during the early design and renovation phases of buildings, training data are almost entirely simulated. Therefore, using methods like feature selection and principal component analysis to choose highly correlated input parameters is crucial for obtaining accurate energy consumption simulations, which in turn enhances the accuracy of prediction models. Future research should focus on continuously refining and processing data to improve their quality and reliability, thereby providing a stronger foundation for subsequent modeling and analysis.

4.2. Optimization of Model Parameters and Algorithms

Model parameter optimization is pivotal in predicting building energy consumption. Investigating automated optimization techniques, such as hyperparameter tuning, can greatly improve prediction model accuracy by refining neural network structures and parameters. Numerous studies have already boosted prediction accuracy by employing optimization methods like singular spectrum analysis [177], random search [237], exhaustive grid search [88], Taguchi methods [240], Bayesian optimization [68,87,92,114,116,141,144,241], and genetic algorithms [64,66,108,115,120,136,139,190,242]. These techniques enable more effective exploration of the parameter space, optimizing essential parameters and designing robust experiments to find the best parameter combinations, thereby significantly enhancing model performance and predictive accuracy. Future developments should focus on optimizing neural network structures and incorporating online learning and dynamic adjustments to further improve predictive outcomes and functionality.
While traditional algorithms encounter numerous challenges such as limited data volume, complex models, and poor generalization, the building energy consumption prediction domain is in dire need of algorithmic advancements to boost model accuracy and robustness. Contemporary research has incorporated various optimization algorithms like genetic algorithms [64,66,108,115,120,136,139,190,242], population-based incremental learning [220], evolutionary strategies [220], particle swarm optimization [45,105,164,219,220,228,243,244], ant colony optimization [176,220], electromagnetic firefly algorithm [46], biogeography-based optimization [220], TLBO [53,83,235], and GWO [141,180,228] with artificial neural networks. These algorithms are instrumental in optimizing parameters, structuring, accelerating training, and enhancing neural network performance, helping them to better accommodate diverse data and tasks, thereby augmenting predictive power and generalization. Additionally, advancements in regularization methods like dropout [142,144] and optimization algorithms such as iPSO [115,154] and iTLBO [52] have substantially improved the efficiency and effectiveness of neural networks.
An increasing number of studies are now shifting from trial-and-error methods to algorithm optimization for setting model parameters. However, traditional algorithms often lack sufficient generalizability. Future research should employ improved algorithms, selecting the most suitable ones based on the specific requirements of the problem and the characteristics of the data. This approach will enhance generalizability and efficiency, improve model interpretability and explainability, and ultimately increase the accuracy of building energy consumption predictions.

4.3. Applications of Integrated and Hybrid Models

In the process of constructing neural network models, the choice of neural network architecture is crucial. The previous sections proposed trends in selecting different neural network structures at different building phases, noting an increasing application of ensemble learning and hybrid models. A substantial body of literature now exists [204,205,206,207,208,209,210,211,212,213,214,215] on predicting building energy consumption using ensemble learning methods, as well as numerous studies [19,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203] using hybrid models for precise energy consumption predictions. Results demonstrate that ensemble machine learning models possess good predictive accuracy in forecasting building energy consumption.
Research has found that in the domain of building energy consumption prediction, ensemble learning is an effective approach. Ensemble learning, which combines multiple base models to enhance predictive performance and generalization, includes methods such as random forests, gradient boosting trees, adaptive boosting, and stacked ensembling. Utilizing ensemble learning techniques can yield more accurate results for building energy consumption prediction. Hybrid models, which combine different types of models to enhance predictive performance and robustness, are also employed in building energy consumption prediction. Hybrid models can integrate traditional statistical models (such as linear regression and logistic regression) with neural networks, leveraging the strengths of both to improve prediction accuracy and generalization. They can also combine physics-based models with ensemble and deep learning approaches, utilizing both the physical properties of buildings and extensive empirical data for precise predictions of building energy consumption.
Research in the operational phase of buildings often requires extensive real-world data, making the application of ensemble learning and hybrid models particularly impactful during this stage. These methods can significantly enhance prediction performance. In the future, as building data and models continue to accumulate, ensemble learning and hybrid models will play an increasingly vital role. The development of these approaches may focus on enhancing the diversity in model combinations, dynamic weight adjustment and adaptive learning, integration of heterogeneous models, combining ensemble learning with deep learning, applying online and incremental learning, and improving model interpretability and transparency. These advancements will further improve the accuracy, stability, and interpretability of building energy consumption predictions, driving the development and application of energy consumption prediction technologies.

5. Conclusions

A comprehensive review of the literature from 2015 to 2023, focusing on 292 papers, summarizes the characteristics of different ANN structures employed in building energy consumption prediction. By analyzing these structures and their practical applications in the design, operation, and renovation phases of buildings, this paper not only highlights the comprehensive application of various input types, building types, energy types, and temporal characteristics in energy consumption prediction models, but also proposes how to select appropriate ANN structures based on different building phases.
In the early design phase, fully connected feedforward neural networks (FFNN) are popular due to their simplicity and efficiency, while convolutional neural networks (CNN) are popular due to their efficiency in processing spatial data. During the operational phase, recurrent neural networks (RNNs), particularly long short-term memory networks (LSTMs), are predominant for their ability to handle time series data, leading to improved prediction accuracy; CNNs and hybrid models are also increasingly used for their strong capabilities in managing complex data structures. For the building renovation phase, feedforward neural networks are preferred due to their usability, quick training, and broad applicability, allowing for swift evaluation and adjustments to building performance.
Based on the literature review and analysis, future research directions may include the following:
  • Exploring improved optimization algorithms to enhance the performance of artificial neural networks;
  • Exploring optimization of neural network structures and parameter settings to enable online learning and dynamic adjustments to enhance prediction outcomes;
  • Investigating the application of ensemble learning and hybrid models for dynamic weight adjustment and adaptive learning, integration of heterogeneous models, online and incremental learning, and enhancing model interpretability and transparency;
  • Exploring more effective data cleaning, feature engineering, and data integration methods to improve data quality and model predictive performance.

Author Contributions

Conceptualization, Q.Y.; methodology, C.H.; software, C.H.; validation, Y.L.; resources, A.L.; data curation, X.L.; writing—original draft preparation, Q.Y. and C.H.; writing—review and editing, Q.Y. and C.H; and project administration, C.H. All authors have read and agreed to the published version of the manuscript.

Funding

Qing Yin reports financial support was provided by the National Natural Science Foundation of China for the project titled “Research on Low-Energy Residential Design in Cold Region Villages and Towns Combining Dynamic Building Energy Consumption Simulation and Thermal Experiments” (Grant No. 52078155).

Acknowledgments

I would like to thank my master’s supervisor, Associate Professor Yin Qing, for his valuable advice and guidance on research methods. I would also like to thank Professor Ying Liu of Harbin Institute of Technology for her professional guidance in writing this dissertation.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Ahmad, M.W.; Mourshed, M.; Rezgui, Y. Trees vs Neurons: Comparison between Random Forest and ANN for High-Resolution Prediction of Building Energy Consumption. Energy Build. 2017, 147, 77–89. [Google Scholar] [CrossRef]
  2. Roman, N.D.; Bre, F.; Fachinotti, V.D.; Lamberts, R. Application and Characterization of Metamodels Based on Artificial Neural Networks for Building Performance Simulation: A Systematic Review. Energy Build. 2020, 217, 109972. [Google Scholar] [CrossRef]
  3. Lu, C.; Li, S.; Lu, Z. Building Energy Prediction Using Artificial Neural Networks: A Literature Survey. Energy Build. 2022, 262, 111718. [Google Scholar] [CrossRef]
  4. Guyot, D.; Giraud, F.; Simon, F.; Corgier, D.; Marvillet, C.; Tremeac, B. Overview of the Use of Artificial Neural Networks for Energy-Related Applications in the Building Sector. Int. J. Energy Res. 2019, 43, 6680–6720. [Google Scholar] [CrossRef]
  5. Md Ramli, S.S.; Nizam Ibrahim, M.; Mohamad, A.; Daud, K.; Saidina Omar, A.M.; Ahmad, N.D. Review of Artificial Neural Network Approaches for Predicting Building Energy Consumption. In Proceedings of the 2023 IEEE 3rd International Conference in Power Engineering Applications (ICPEA), Putrajaya, Malaysia, 6–7 March 2023; pp. 328–333. [Google Scholar]
  6. Mohandes, S.; Zhang, X.; Mahdiyar, A. A Comprehensive Review on the Application of Artificial Neural Networks in Building Energy Analysis. Neurocomputing 2019, 340, 55–75. [Google Scholar] [CrossRef]
  7. Runge, J.; Zmeureanu, R. Forecasting Energy Use in Buildings Using Artificial Neural Networks: A Review. Energies 2019, 12, 3254. [Google Scholar] [CrossRef]
  8. Widrow, B. Adaptive “Adaline” Neuron Using Chemical “Memistors”; Stanford Electronics Laboratories: Stanford, CA, USA, 1960. [Google Scholar]
  9. Islam, S.M.; Al-Alawi, S.M.; Ellithy, K.A. Forecasting Monthly Electric Load and Energy for a Fast Growing Utility Using an Artificial Neural Network. Electr. Power Syst. Res. 1995, 34, 1–9. [Google Scholar] [CrossRef]
  10. Kalogirou, S.A. Applications of Artificial Neural-Networks for Energy Systems. Appl. Energy 2000, 67, 17–35. [Google Scholar] [CrossRef]
  11. González, P.A.; Zamarreño, J.M. Prediction of Hourly Energy Consumption in Buildings Based on a Feedback Artificial Neural Network. Energy Build. 2005, 37, 595–601. [Google Scholar] [CrossRef]
  12. Karatasou, S.; Santamouris, M.; Geros, V. Modeling and Predicting Building’s Energy Use with Artificial Neural Networks: Methods and Results. Energy Build. 2006, 38, 949–958. [Google Scholar] [CrossRef]
  13. Marino, D.L.; Amarasinghe, K.; Manic, M. Building Energy Load Forecasting Using Deep Neural Networks. In Proceedings of the IECON 2016—42nd Annual Conference of the IEEE Industrial Electronics Society, Florence, Italy, 24–27 October 2016; pp. 7046–7051. [Google Scholar]
  14. Shi, H.; Xu, M.; Li, R. Deep Learning for Household Load Forecasting—A Novel Pooling Deep RNN. IEEE Trans. Smart Grid 2018, 9, 5271–5280. [Google Scholar] [CrossRef]
  15. Rahman, A.; Srikumar, V.; Smith, A.D. Predicting Electricity Consumption for Commercial and Residential Buildings Using Deep Recurrent Neural Networks. Appl. Energy 2018, 212, 372–385. [Google Scholar] [CrossRef]
  16. Kumar, S.; Hussain, L.; Banarjee, S.; Reza, M. Energy Load Forecasting Using Deep Learning Approach-LSTM and GRU in Spark Cluster. In Proceedings of the 2018 Fifth International Conference on Emerging Applications of Information Technology (EAIT), Kolkata, India, 12–13 January 2018; IEEE: Piscataway, NJ, USA, 2018; pp. 1–4. [Google Scholar]
  17. Kim, T.Y.; Cho, S.B. Predicting Residential Energy Consumption Using CNN-LSTM Neural Networks. Energy 2019, 182, 72–81. [Google Scholar] [CrossRef]
  18. Gao, Y.; Ruan, Y.; Fang, C.; Yin, S. Deep Learning and Transfer Learning Models of Energy Consumption Forecasting for a Building with Poor Information Data. Energy Build. 2020, 223, 110156. [Google Scholar] [CrossRef]
  19. Li, G.; Zhao, X.; Fan, C.; Fang, X.; Li, F.; Wu, Y. Assessment of Long Short-Term Memory and Its Modifications for Enhanced Short-Term Building Energy Predictions. J. Build. Eng. 2021, 43, 103182. [Google Scholar] [CrossRef]
  20. Aliyu, I.; Um, T.W.; Lee, S.J.; Lim, C.; Kim, J. Deep Learning for Multivariate Prediction of Building Energy Performance of Residential Buildings. Comput. Mater. Contin. 2023, 75, 5947–5964. [Google Scholar] [CrossRef]
  21. Singaravel, S.; Suykens, J.; Geyer, P. Deep Convolutional Learning for General Early Design Stage Prediction Models. Adv. Eng. Inform. 2019, 42, 100982. [Google Scholar] [CrossRef]
  22. Olu-Ajayi, R.; Alaka, H.; Sulaimon, I.; Sunmola, F.; Ajayi, S. Machine Learning for Energy Performance Prediction at the Design Stage of Buildings. Energy Sustain. Dev. 2022, 66, 12–25. [Google Scholar] [CrossRef]
  23. Elbeltagi, E.; Wefki, H. Predicting Energy Consumption for Residential Buildings Using ANN through Parametric Modeling. Energy Rep. 2021, 7, 2534–2545. [Google Scholar] [CrossRef]
  24. Singh, M.M.; Smith, I.F.C. Convolutional Neural Network to Learn Building-Shape Representations for Early-Stage Energy Design. Energy AI 2023, 14, 100293. [Google Scholar] [CrossRef]
  25. Himmetoğlu, S.; Delice, Y.; Kızılkaya Aydoğan, E.; Uzal, B. Green Building Envelope Designs in Different Climate and Seismic Zones: Multi-Objective ANN-Based Genetic Algorithm. Sustain. Energy Technol. Assess. 2022, 53, 102505. [Google Scholar] [CrossRef]
  26. Sapnken, F.E.; Hamed, M.M.; Soldo, B.; Tamba, J.G. Modeling Energy-Efficient Building Loads Using Machine-Learning Algorithms for the Design Phase. Energy Build. 2023, 283, 112807. [Google Scholar] [CrossRef]
  27. Li, Z.; Dai, J.; Chen, H.; Lin, B. An ANN-Based Fast Building Energy Consumption Prediction Method for Complex Architectural Form at the Early Design Stage. Build. Simul. 2019, 12, 665–681. [Google Scholar] [CrossRef]
  28. Li, X.; Liu, S.; Zhao, L.; Meng, X.; Fang, Y. An Integrated Building Energy Performance Evaluation Method: From Parametric Modeling to GA-NN Based Energy Consumption Prediction Modeling. J. Build. Eng. 2022, 45, 103571. [Google Scholar] [CrossRef]
  29. Guo, H.; Duan, D.; Yan, J.; Ding, K.; Xiang, F.; Peng, R. Machine Learning-Based Method for Detached Energy-Saving Residential Form Generation. Buildings 2022, 12, 1504. [Google Scholar] [CrossRef]
  30. Wang, M.; Cao, S.; Chen, D.; Ji, G.; Ma, Q.; Ren, Y. Research on Design Framework of Middle School Teaching Building Based on Performance Optimization and Prediction in the Scheme Design Stage. Buildings 2022, 12, 1897. [Google Scholar] [CrossRef]
  31. Chari, A.; Christodoulou, S. Building Energy Performance Prediction Using Neural Networks. Energy Effic. 2017, 10, 1315–1327. [Google Scholar] [CrossRef]
  32. Jihad, A.S.; Tahiri, M. Forecasting the Heating and Cooling Load of Residential Buildings by Using a Learning Algorithm “Gradient Descent”, Morocco. Case Stud. Therm. Eng. 2018, 12, 85–93. [Google Scholar] [CrossRef]
  33. Moayedi, H.; Nguyen, H.; Kok Foong, L. Nonlinear Evolutionary Swarm Intelligence of Grasshopper Optimization Algorithm and Gray Wolf Optimization for Weight Adjustment of Neural Network. Eng. Comput. 2021, 37, 1265–1275. [Google Scholar] [CrossRef]
  34. Martellotta, F.; Ayr, U.; Stefanizzi, P.; Sacchetti, A.; Riganti, G.; Fortunato, B.; Ficarella, A.; Torresi, M. On the Use of Artificial Neural Networks to Model Household Energy Consumptions. Energy Procedia 2017, 126, 250–257. [Google Scholar] [CrossRef]
  35. Xu, X.; Feng, G.; Chi, D.; Liu, M.; Dou, B. Optimization of Performance Parameter Design and Energy Use Prediction for Nearly Zero Energy Buildings. Energies 2018, 11, 3252. [Google Scholar] [CrossRef]
  36. Moradzadeh, A.; Mansour-Saatloo, A.; Mohammadi-Ivatloo, B.; Anvari-Moghaddam, A. Performance Evaluation of Two Machine Learning Techniques in Heating and Cooling Loads Forecasting of Residential Buildings. Appl. Sci. 2020, 10, 3829. [Google Scholar] [CrossRef]
  37. Moayedi, H.; Mosavi, A. Suggesting a Stochastic Fractal Search Paradigm in Combination with Artificial Neural Network for Early Prediction of Cooling Load in Residential Buildings. Energies 2021, 14, 1649. [Google Scholar] [CrossRef]
  38. Wu, D.; Foong, L.K.; Lyu, Z. Two Neural-Metaheuristic Techniques Based on Vortex Search and Backtracking Search Algorithms for Predicting the Heating Load of Residential Buildings. Eng. Comput. 2022, 38, 647–660. [Google Scholar] [CrossRef]
  39. Zhou, Y.; Liang, Y.; Pan, Y.; Yuan, X.; Xie, Y.; Jia, W. A Deep-Learning-Based Meta-Modeling Workflow for Thermal Load Forecasting in Buildings: Method and a Case Study. Buildings 2022, 12, 177. [Google Scholar] [CrossRef]
  40. Gong, Y.; Zoltán, E.S.; Gyergyák, J. A Neural Network Trained by Multi-Tracker Optimization Algorithm Applied to Energy Performance Estimation of Residential Buildings. Buildings 2023, 13, 1167. [Google Scholar] [CrossRef]
  41. Uba, F.; Apevienyeku, H.K.; Nsiah, F.D.; Akorli, A.; Adjignon, S. Energy Analysis of Commercial Buildings Using Artificial Neural Network. Model. Simul. Eng. 2021, 2021, e8897443. [Google Scholar] [CrossRef]
  42. Zhussupbekov, M.; Memon, S.A.; Khawaja, S.A.; Nazir, K.; Kim, J. Forecasting Energy Demand of PCM Integrated Residential Buildings: A Machine Learning Approach. J. Build. Eng. 2023, 70, 106335. [Google Scholar] [CrossRef]
  43. Senarathne, L.R.; Nanda, G.; Sundararajan, R. Influence of Building Parameters on Energy Efficiency Levels: A Bayesian Network Study. Adv. Build. Energy Res. 2022, 16, 780–805. [Google Scholar] [CrossRef]
  44. Pachauri, N.; Ahn, C.W. Weighted Aggregated Ensemble Model for Energy Demand Management of Buildings. Energy 2023, 263, 125853. [Google Scholar] [CrossRef]
  45. Le, L.T.; Nguyen, H.; Dou, J.; Zhou, J. A Comparative Study of PSO-ANN, GA-ANN, ICA-ANN, and ABC-ANN in Estimating the Heating Load of Buildings’ Energy Efficiency for Smart City Planning. Appl. Sci. 2019, 9, 2630. [Google Scholar] [CrossRef]
  46. Bui, D.K.; Nguyen, T.N.; Ngo, T.D.; Nguyen-Xuan, H. An Artificial Neural Network (ANN) Expert System Enhanced with the Electromagnetism-Based Firefly Algorithm (EFA) for Predicting the Energy Consumption in Buildings. Energy 2020, 190, 116370. [Google Scholar] [CrossRef]
  47. Luo, X.J.; Oyedele, L.O.; Ajayi, A.O.; Akinade, O. Comparative Study of Machine Learning-Based Multi-Objective Prediction Framework for Multiple Building Energy Loads. Sustain. Cities Soc. 2020, 61, 102283. [Google Scholar] [CrossRef]
  48. Lei, L.; Chen, W.; Wu, B.; Chen, C.; Liu, W. A Building Energy Consumption Prediction Model Based on Rough Set Theory and Deep Learning Algorithms. Energy Build. 2021, 240, 110886. [Google Scholar] [CrossRef]
  49. Moon, J.; Park, S.; Rho, S.; Hwang, E. A Comparative Analysis of Artificial Neural Network Architectures for Building Energy Consumption Forecasting. Int. J. Distrib. Sens. Netw. 2019, 15, 1550147719877616. [Google Scholar] [CrossRef]
  50. Fayaz, M.; Shah, H.; Aseere, A.; Mashwani, W.; Shah, A. A Framework for Prediction of Household Energy Consumption Using Feed Forward Back Propagation Neural Network. Technologies 2019, 7, 30. [Google Scholar] [CrossRef]
  51. Zhang, C.; Zhao, Y.; Fan, C.; Li, T.; Zhang, X.; Li, J. A Generic Prediction Interval Estimation Method for Quantifying the Uncertainties in Ultra-Short-Term Building Cooling Load Prediction. Appl. Therm. Eng. 2020, 173, 115261. [Google Scholar] [CrossRef]
  52. Li, K.; Xie, X.; Xue, W.; Dai, X.; Chen, X.; Yang, X. A Hybrid Teaching-Learning Artificial Neural Network for Building Electrical Energy Consumption Prediction. Energy Build. 2018, 174, 323–334. [Google Scholar] [CrossRef]
  53. Almutairi, K.; Algarni, S.; Alqahtani, T.; Moayedi, H.; Mosavi, A. A TLBO-Tuned Neural Processor for Predicting Heating Load in Residential Buildings. Sustainability 2022, 14, 5924. [Google Scholar] [CrossRef]
  54. Walker, S.; Khan, W.; Katic, K.; Maassen, W.; Zeiler, W. Accuracy of Different Machine Learning Algorithms and Added-Value of Predicting Aggregated-Level Energy Performance of Commercial Buildings. Energy Build. 2020, 209, 109705. [Google Scholar] [CrossRef]
  55. Truong, L.; Chow, K.; Luevisadpaibul, R.; Thirunavukkarasu, G.; Seyedmahmoudian, M.; Horan, B.; Mekhilef, S.; Stojcevski, A. Accurate Prediction of Hourly Energy Consumption in a Residential Building Based on the Occupancy Rate Using Machine Learning Approaches. Appl. Sci. 2021, 11, 2229. [Google Scholar] [CrossRef]
  56. Ling, J.; Dai, N.; Xing, J.; Tong, H. An Improved Input Variable Selection Method of the Data-Driven Model for Building Heating Load Prediction. J. Build. Eng. 2021, 44, 103255. [Google Scholar] [CrossRef]
  57. Ciulla, G.; D’Amico, A.; Lo Brano, V.; Traverso, M. Application of Optimized Artificial Intelligence Algorithm to Evaluate the Heating Energy Demand of Non-Residential Buildings at European Level. Energy 2019, 176, 380–391. [Google Scholar] [CrossRef]
  58. Kim, M.; Jung, S.; Kang, J. Artificial Neural Network-Based Residential Energy Consumption Prediction Models Considering Residential Building Information and User Features in South Korea. Sustainability 2020, 12, 109. [Google Scholar] [CrossRef]
  59. Son, N.; Yang, S.; Na, J. Deep Neural Network and Long Short-Term Memory for Electric Power Load Forecasting. Appl. Sci. 2020, 10, 6489. [Google Scholar] [CrossRef]
  60. Li, A.; Xiao, F.; Fan, C.; Hu, M. Development of an ANN-Based Building Energy Model for Information-Poor Buildings Using Transfer Learning. Build. Simul. 2021, 14, 89–101. [Google Scholar] [CrossRef]
  61. Jang, J.; Lee, J.; Son, E.; Park, K.; Kim, G.; Lee, J.; Leigh, S. Development of an Improved Model to Predict Building Thermal Energy Consumption by Utilizing Feature Selection. Energies 2019, 12, 4187. [Google Scholar] [CrossRef]
  62. Kampelis, N.; Tsekeri, E.; Kolokotsa, D.; Kalaitzakis, K.; Isidori, D.; Cristalli, C. Development of Demand Response Energy Management Optimization at Building and District Levels Using Genetic Algorithm and Artificial Neural Network Modelling Power Predictions. Energies 2018, 11, 3012. [Google Scholar] [CrossRef]
  63. Benedetti, M.; Cesarotti, V.; Introna, V.; Serranti, J. Energy Consumption Control Automation Using Artificial Neural Networks and Adaptive Algorithms: Proposal of a New Methodology and Case Study. Appl. Energy 2016, 165, 60–71. [Google Scholar] [CrossRef]
  64. Luo, X.; Oyedele, L.; Ajayi, A.; Akinade, O.; Owolabi, H.; Ahmed, A. Feature Extraction and Genetic Algorithm Enhanced Adaptive Deep Neural Network for Energy Consumption Prediction in Buildings. Renew. Sustain. Energy Rev. 2020, 131, 109980. [Google Scholar] [CrossRef]
  65. Deb, C.; Eang, L.; Yang, J.; Santamouris, M.; Sun, Y.; Pei, Y. Forecasting Energy Consumption of Institutional Buildings in Singapore. Procedia Eng. 2015, 121, 1734–1740. [Google Scholar] [CrossRef]
  66. Luo, X.; Oyedele, L.; Ajayi, A.; Akinade, O.; Delgado, J.; Owolabi, H.; Ahmed, A. Genetic Algorithm-Determined Deep Feedforward Neural Network Architecture for Predicting Electricity Consumption in Real Buildings. Energy AI 2020, 2, 100015. [Google Scholar] [CrossRef]
  67. Dong, Z.; Liu, J.; Liu, B.; Li, L.; Li, X. Hourly Energy Consumption Prediction of an Office Building Based on Ensemble Learning and Energy Consumption Pattern Classification. Energy Build. 2021, 241, 110929. [Google Scholar] [CrossRef]
  68. Kim, M.; Kim, Y.; Srebric, J. Impact of Correlation of Plug Load Data, Occupancy Rates and Local Weather Conditions on Electricity Consumption in a Building Using Four Back-Propagation Neural Network Models. Sustain. Cities Soc. 2020, 62, 102321. [Google Scholar] [CrossRef]
  69. Xu, X.; Wang, W.; Hong, T.; Chen, J. Incorporating Machine Learning with Building Network Analysis to Predict Multi-Building Energy Use. Energy Build. 2019, 186, 80–97. [Google Scholar] [CrossRef]
  70. Gunay, B.; Shen, W.; Newsham, G. Inverse Blackbox Modeling of the Heating and Cooling Load in Office Buildings. Energy Build. 2017, 142, 200–210. [Google Scholar] [CrossRef]
  71. Carrera, B.; Peyrard, S.; Kim, K. Meta-Regression Framework for Energy Consumption Prediction in a Smart City: A Case Study of Songdo in South Korea. Sustain. Cities Soc. 2021, 72, 103025. [Google Scholar] [CrossRef]
  72. Dagdougui, H.; Bagheri, F.; Le, H.; Dessaint, L. Neural Network Model for Short-Term and Very-Short-Term Load Forecasting in District Buildings. Energy Build. 2019, 203, 109408. [Google Scholar] [CrossRef]
  73. Wang, L.; Lee, E.; Yuen, R. Novel Dynamic Forecasting Model for Building Cooling Loads Combining an Artificial Neural Network and an Ensemble Approach. Appl. Energy 2018, 228, 1740–1753. [Google Scholar] [CrossRef]
  74. Anand, P.; Deb, C.; Yan, K.; Yang, J.; Cheong, D.; Sekhar, C. Occupancy-Based Energy Consumption Modelling Using Machine Learning Algorithms for Institutional Buildings. Energy Build. 2021, 252, 111478. [Google Scholar] [CrossRef]
  75. Ye, Z.; Kim, M. Predicting Electricity Consumption in a Building Using an Optimized Back-Propagation and Levenberg-Marquardt Back-Propagation Neural Network: Case Study of a Shopping Mall in China. Sustain. Cities Soc. 2018, 42, 176–183. [Google Scholar] [CrossRef]
  76. Lee, S.; Jung, S.; Lee, J. Prediction Model Based on an Artificial Neural Network for User-Based Building Energy Consumption in South Korea. Energies 2019, 12, 608. [Google Scholar] [CrossRef]
  77. Jang, J.; Baek, J.; Leigh, S. Prediction of Optimum Heating Timing Based on Artificial Neural Network by Utilizing BEMS Data. J. Build. Eng. 2019, 22, 66–74. [Google Scholar] [CrossRef]
  78. Alshibani, A. Prediction of the Energy Consumption of School Buildings. Appl. Sci. 2020, 10, 5885. [Google Scholar] [CrossRef]
  79. Kim, M.; Kim, Y.; Srebric, J. Predictions of Electricity Consumption in a Campus Building Using Occupant Rates and Weather Elements with Sensitivity Analysis: Artificial Neural Network vs. Linear Regression. Sustain. Cities Soc. 2020, 62, 102385. [Google Scholar] [CrossRef]
  80. Zhang, C.; Li, J.; Zhao, Y.; Li, T.; Chen, Q.; Zhang, X.; Qiu, W. Problem of Data Imbalance in Building Energy Load Prediction: Concept, Influence, and Solution. Appl. Energy 2021, 297, 117139. [Google Scholar] [CrossRef]
  81. Kassaymeh, S.; Abdullah, S.; Al-Laham, M.; Alweshah, M.; Al-Betar, M.; Othman, Z. Salp Swarm Optimizer for Modeling Software Reliability Prediction Problems. Neural Process. Lett. 2021, 53, 4451–4487. [Google Scholar] [CrossRef]
  82. Kim, M.; Cha, J.; Lee, E.; Pham, V.; Lee, S.; Theera-Umpon, N. Simplified Neural Network Model Design with Sensitivity Analysis and Electricity Consumption Prediction in a Commercial Building. Energies 2019, 12, 1201. [Google Scholar] [CrossRef]
  83. Zhou, G.; Moayedi, H.; Foong, L.K. Teaching–Learning-Based Metaheuristic Scheme for Modifying Neural Computing in Appraising Energy Performance of Building. Eng. Comput. 2021, 37, 3037–3048. [Google Scholar] [CrossRef]
  84. Macas, M.; Moretti, F.; Fonti, A.; Giantomassi, A.; Comodi, G.; Annunziato, M.; Pizzuti, S.; Capra, A. The Role of Data Sample Size and Dimensionality in Neural Network Based Forecasting of Building Heating Related Variables. Energy Build. 2016, 111, 299–310. [Google Scholar] [CrossRef]
  85. Hu, J.; Zheng, W.; Zhang, S.; Li, H.; Liu, Z.; Zhang, G.; Yang, X. Thermal Load Prediction and Operation Optimization of Office Building with a Zone-Level Artificial Neural Network and Rule-Based Control. Appl. Energy 2021, 300, 117429. [Google Scholar] [CrossRef]
  86. Deb, C.; Lee, S.; Santamouris, M. Using Artificial Neural Networks to Assess HVAC Related Energy Saving in Retrofitted Office Buildings. SOLAR Energy 2018, 163, 32–44. [Google Scholar] [CrossRef]
  87. Bampoulas, A.; Pallonetto, F.; Mangina, E.; Finn, D. A Bayesian Deep-Learning Framework for Assessing the Energy Flexibility of Residential Buildings with Multicomponent Energy Systems. Appl. Energy 2023, 348, 1–25. [Google Scholar] [CrossRef]
  88. Yue, N.; Caini, M.; Li, L.; Zhao, Y.; Li, Y. A Comparison of Six Metamodeling Techniques Applied to Multi Building Performance Vectors Prediction on Gymnasiums under Multiple Climate Conditions. Appl. Energy 2023, 332, 120481. [Google Scholar] [CrossRef]
  89. Elshaboury, N.; Abdelkader, E.; Al-Sakkaf, A.; Bagchi, A. A Deep Convolutional Neural Network for Predicting Electricity Consumption at Grey Nuns Building in Canada. Constr. Innov. 2023; ahead of print. [Google Scholar]
  90. Jing, W.; Zhen, M.; Guan, H.; Luo, W.; Liu, X. A Prediction Model for Building Energy Consumption in a Shopping Mall Based on Chaos Theory. Energy Rep. 2022, 8, 5305–5312. [Google Scholar] [CrossRef]
  91. Manno, A.; Martelli, E.; Amaldi, E. A Shallow Neural Network Approach for the Short-Term Forecast of Hourly Energy Consumption. Energies 2022, 15, 958. [Google Scholar] [CrossRef]
  92. Zheng, S.; Zhong, Q.; Peng, L.; Chai, X. A Simple Method of Residential Electricity Load Forecasting by Improved Bayesian Neural Networks. Math. Probl. Eng. 2018, 2018, 4276176. [Google Scholar] [CrossRef]
  93. Rastbod, S.; Rahimi, F.; Dehghan, Y.; Kamranfar, S.; Benjeddou, O.; Nehdi, M. An Optimized Machine Learning Approach for Forecasting Thermal Energy Demand of Buildings. Sustainability 2023, 15, 231. [Google Scholar] [CrossRef]
  94. Erdem, M.; Sariçiçek, O.; Calis, G. Benchmarking Performance of Machine-Learning Methods for Building Energy Demand Modelling. In Proceedings of the Institution of Civil Engineers-Engineering Sustainability; Thomas Telford Ltd.: London, UK, 2022. [Google Scholar]
  95. Seo, B.; Yoon, Y.; Lee, K.; Cho, S. Comparative Analysis of ANN and LSTM Prediction Accuracy and Cooling Energy Savings through AHU-DAT Control in an Office Building. Buildings 2023, 13, 1434. [Google Scholar] [CrossRef]
  96. Wang, S.; Ma, H.; Dou, J.; Ye, Y.; Yang, J. Energy Consumption Prediction of Public Buildings Based on MEA-BP Neural Network. In Proceedings of the 2021 33rd Chinese Control and Decision Conference (CCDC), Kunming, China, 22–24 May 2021; pp. 850–855. [Google Scholar]
  97. Truong, N.; Ngo, N.; Pham, A. Forecasting Time-Series Energy Data in Buildings Using an Additive Artificial Intelligence Model for Improving Energy Efficiency. Comput. Intell. Neurosci. 2021, 2021, 6028573. [Google Scholar] [CrossRef]
  98. Das, S.; Swetapadma, A.; Panigrahi, C.; Abdelaziz, A. Improved Method for Approximation of Heating and Cooling Load in Urban Buildings for Energy Performance Enhancement. Electr. Power Compon. Syst. 2020, 48, 436–446. [Google Scholar] [CrossRef]
  99. Cordeiro-Costas, M.; Villanueva, D.; Eguía-Oller, P.; Martínez-Comesaña, M.; Ramos, S. Load Forecasting with Machine Learning and Deep Learning Methods. Appl. Sci. 2023, 13, 7933. [Google Scholar] [CrossRef]
  100. Alotaibi, M. Machine Learning Approach for Short-Term Load Forecasting Using Deep Neural Network. Energies 2022, 15, 6261. [Google Scholar] [CrossRef]
  101. Yu, M.; Li, L.; Guo, Z. Model Analysis of Energy Consumption Data for Green Building Using Deep Learning Neural Network. Int. J. Low Carbon Technol. 2022, 17, 233–244. [Google Scholar] [CrossRef]
  102. Wang, H.; Li, H.; Gao, S.; Zhou, Z.; Lao, Z. Multigranularity Building Energy Consumption Prediction Method Based on Convolutional Recurrent Neural Network. Wirel. Commun. Mob. Comput. 2022, 2022, 8524034. [Google Scholar] [CrossRef]
  103. Lee, S.; Cha, J.; Kim, M.K.; Kim, K.S.; Pham, V.H.; Leach, M. Neural-Network-Based Building Energy Consumption Prediction with Training Data Generation. Processes 2019, 7, 731. [Google Scholar] [CrossRef]
  104. Fallah, A.; Ghafourian, E.; Sichani, L.; Ghafourian, H.; Arandian, B.; Nehdi, M. Novel Neural Network Optimized by Electrostatic Discharge Algorithm for Modification of Buildings Energy Performance. Sustainability 2023, 15, 2884. [Google Scholar] [CrossRef]
  105. Mustapa, R.; Dahlan, N.; Yassin, I.; Nordin, A.; Mahadan, M.; Mohamad, S. Particle Swarm Optimization for NARX-ANN Baseline Energy Modelling in Educational Buildings. In Proceedings of the 2019 IEEE International Conference on Automatic Control and Intelligent Systems (I2CACIS), Selangor, Malaysia, 29 June 2019; pp. 19–24. [Google Scholar]
  106. Lei, R.; Yin, J. Prediction Method of Energy Consumption for High Building Based on LMBP Neural Network. Energy Rep. 2022, 8, 1236–1248. [Google Scholar] [CrossRef]
  107. Cai, D.; Duan, P.; Li, J. Prediction of Air Conditioning Energy Consumption Based on BP Neural Networks with an Each-Column Optimization Synthesis Algorithm. In Proceedings of the 2021 IEEE 16th Conference on Industrial Electronics and Applications (ICIEA), Chengdu, China, 1–4 August 2021; pp. 1696–1701. [Google Scholar]
  108. Lu, Y.; Gong, X.; Kipnis, A. Prediction of Low-Energy Building Energy Consumption Based on Genetic BP Algorithm. CMC-Comput. Mater. Contin. 2022, 72, 5481–5497. [Google Scholar] [CrossRef]
  109. Lee, C.; Jung, D.; Lee, D.; Kim, K.; Do, S. Prediction Performance Analysis of Artificial Neural Network Model by Input Variable Combination for Residential Heating Loads. Energies 2021, 14, 756. [Google Scholar] [CrossRef]
  110. Zhang, X.; Zhang, J.; Zhang, J.; Zhang, Y. Research on the Combined Prediction Model of Residential Building Energy Consumption Based on Random Forest and BP Neural Network. Geofluids 2021, 2021, 7271383. [Google Scholar] [CrossRef]
  111. Zhou, C.; Yao, Z.; Hu, Y.; Cui, W. Study On The Application Of BP Neural Network In The Prediction Of Office Building Energy Consumption. In IOP Conference Series: Earth and Environmental Science; IOP Publishing: Bristol, UK, 2020; Volume 546. [Google Scholar]
  112. Dong, B.; Li, Z.; Rahman, S.; Vega, R. A Hybrid Model Approach for Forecasting Future Residential Electricity Consumption. Energy Build. 2016, 117, 341–351. [Google Scholar] [CrossRef]
  113. Yildiz, B.; Bilbao, J.; Sproul, A. A Review and Analysis of Regression and Machine Learning Models on Commercial Building Electricity Load Forecasting. Renew. Sustain. Energy Rev. 2017, 73, 1104–1122. [Google Scholar] [CrossRef]
  114. Chae, Y.; Horesh, R.; Hwang, Y.; Lee, W. Artificial Neural Network Model for Forecasting Sub-Hourly Electricity Usage in Commercial Buildings. Energy Build. 2016, 111, 184–194. [Google Scholar] [CrossRef]
  115. Li, K.; Hu, C.; Liu, G.; Xue, W. Building’s Electricity Consumption Prediction Using Optimized Artificial Neural Networks and Principal Component Analysis. Energy Build. 2015, 108, 106–113. [Google Scholar] [CrossRef]
  116. Deb, C.; Eang, L.; Yang, J.; Santamouris, M. Forecasting Diurnal Cooling Energy Load for Institutional Buildings Using Artificial Neural Networks. Energy Build. 2016, 121, 284–297. [Google Scholar] [CrossRef]
  117. Platon, R.; Dehkordi, V.R.; Martel, J. Hourly Prediction of a Building’s Electricity Consumption Using Case-Based Reasoning, Artificial Neural Networks and Principal Component Analysis. Energy Build. 2015, 92, 10–18. [Google Scholar] [CrossRef]
  118. Amber, K.; Ahmad, R.; Aslam, M.; Kousar, A.; Usman, M.; Khan, M. Intelligent Techniques for Forecasting Electricity Consumption of Buildings. Energy 2018, 157, 886–893. [Google Scholar] [CrossRef]
  119. Zulfiqar, M.; Kamran, M.; Rasheed, M.; Alquthami, T.; Milyani, A. A Short-Term Load Forecasting Model Based on Self-Adaptive Momentum Factor and Wavelet Neural Network in Smart Grid. IEEE Access 2022, 10, 77587–77602. [Google Scholar] [CrossRef]
  120. Yaqing, T. Prediction of Building Power Consumption Based on GAWNN. In IOP Conference Series: Materials Science and Engineering; IOP Publishing: Bristol, UK, 2019; Volume 563, p. 042081. [Google Scholar]
  121. Kumar, S.; Pal, S.; Singh, R. A Novel Method Based on Extreme Learning Machine to Predict Heating and Cooling Load through Design and Structural Attributes. Energy Build. 2018, 176, 275–286. [Google Scholar] [CrossRef]
  122. Li, C.; Ding, Z.; Zhao, D.; Yi, J.; Zhang, G. Building Energy Consumption Prediction: An Extreme Deep Learning Approach. Energies 2017, 10, 1525. [Google Scholar] [CrossRef]
  123. Yu, B.; Kim, D.; Cho, H.; Mago, P. A Nonlinear Autoregressive with Exogenous Inputs Artificial Neural Network Model for Building Thermal Load Prediction. J. Energy Resour. Technol. Trans. ASME 2020, 142, 050902. [Google Scholar] [CrossRef]
  124. Mortazavigazar, A.; Wahba, N.; Newsham, P.; Triharta, M.; Zheng, P.; Chen, T.; Rismanchi, B. Application of Artificial Neural Networks for Virtual Energy Assessment. Energies 2021, 14, 8330. [Google Scholar] [CrossRef]
  125. Zapirain, I.; Etxegarai, G.; Hernández, J.; Boussaada, Z.; Aginako, N.; Camblong, H. Short-Term Electricity Consumption Forecasting with NARX, LSTM, and SVR for a Single Building: Small Data Set Approach. Energy Sources Part A Recovery Util. Environ. Eff. 2022, 44, 6898–6908. [Google Scholar] [CrossRef]
  126. Han, Y.; Fan, C.; Geng, Z.; Ma, B.; Cong, D.; Chen, K.; Yu, B. Energy Efficient Building Envelope Using Novel RBF Neural Network Integrated Affinity Propagation. Energy 2020, 209, 118414. [Google Scholar] [CrossRef]
  127. Koschwitz, D.; Frisch, J.; van Treeck, C. Data-Driven Heating and Cooling Load Predictions for Non-Residential Buildings Based on Support Vector Machine Regression and NARX Recurrent Neural Network: A Comparative Study on District Scale. Energy 2018, 165, 134–142. [Google Scholar] [CrossRef]
  128. Ozer, I.; Efe, S.; Ozbay, H. A Combined Deep Learning Application for Short Term Load Forecasting. Alex. Eng. J. 2021, 60, 3807–3818. [Google Scholar] [CrossRef]
  129. Fang, X.; Gong, G.; Li, G.; Chun, L.; Peng, P.; Li, W. A General Multi-Source Ensemble Transfer Learning Framework Integrate of LSTM-DANN and Similarity Metric for Building Energy Prediction. Energy Build. 2021, 252, 111435. [Google Scholar] [CrossRef]
  130. Sendra-Arranz, R.; Gutiérrez, A. A Long Short-Term Memory Artificial Neural Network to Predict Daily HVAC Consumption in Buildings. Energy Build. 2020, 216, 109952. [Google Scholar] [CrossRef]
  131. Gao, Y.; Fang, C.; Ruan, Y. A Novel Model for the Prediction of Long-Term Building Energy Demand: LSTM with Attention Layer. In IOP Conference Series: Earth and Environmental Science; IOP Publishing: Bristol, UK, 2019; Volume 294. [Google Scholar]
  132. Khan, A.; Iqbal, N.; Rizwan, A.; Ahmad, R.; Kim, D. An Ensemble Energy Consumption Forecasting Model Based on Spatial-Temporal Clustering Analysis in Residential Buildings. Energies 2021, 14, 3020. [Google Scholar] [CrossRef]
  133. Li, A.; Xiao, F.; Zhang, C.; Fan, C. Attention-Based Interpretable Neural Network for Building Cooling Load Prediction. Appl. Energy 2021, 299, 117238. [Google Scholar] [CrossRef]
  134. Gao, L.; Liu, T.; Cao, T.; Hwang, Y.; Radermacher, R. Comparing Deep Learning Models for Multi Energy Vectors Prediction on Multiple Types of Building. Appl. Energy 2021, 301, 117486. [Google Scholar] [CrossRef]
  135. Chalapathy, R.; Khoa, N.; Sethuvenkatraman, S. Comparing Multi-Step Ahead Building Cooling Load Prediction Using Shallow Machine Learning and Deep Learning Models. Sustain. Energy Grids Netw. 2021, 28, 100543. [Google Scholar] [CrossRef]
  136. Panagiotou, D.; Dounis, A. Comparison of Hospital Building’s Energy Consumption Prediction Using Artificial Neural Networks, ANFIS, and LSTM Network. Energies 2022, 15, 6453. [Google Scholar] [CrossRef]
  137. Almalaq, A.; Zhang, J. Evolutionary Deep Learning-Based Energy Consumption Prediction for Buildings. IEEE Access 2019, 7, 1520–1531. [Google Scholar] [CrossRef]
  138. Ding, Z.; Chen, W.; Hu, T.; Xu, X. Evolutionary Double Attention-Based Long Short-Term Memory Model for Building Energy Prediction: Case Study of a Green Building. Appl. Energy 2021, 288, 116660. [Google Scholar] [CrossRef]
  139. Luo, X.; Oyedele, L. Forecasting Building Energy Consumption: Adaptive Long-Short Term Memory Neural Networks Driven by Genetic Algorithm. Adv. Eng. Inform. 2021, 50, 101357. [Google Scholar] [CrossRef]
  140. Gao, Y.; Ruan, Y. Interpretable Deep Learning Model for Building Energy Consumption Prediction Based on Attention Mechanism. Energy Build. 2021, 252, 111379. [Google Scholar] [CrossRef]
  141. Lu, Y.; Tian, Z.; Zhou, R.; Liu, W. Multi-Step-Ahead Prediction of Thermal Load in Regional Energy System Using Deep Learning Method. Energy Build. 2021, 233, 110658. [Google Scholar] [CrossRef]
  142. Mahjoub, S.; Chrifi-Alaoui, L.; Marhic, B.; Delahoche, L. Predicting Energy Consumption Using LSTM, Multi-Layer GRU and Drop-GRU Neural Networks. Sensors 2022, 22, 4062. [Google Scholar] [CrossRef]
  143. Ahn, Y.; Kim, B. Prediction of Building Power Consumption Using Transfer Learning-Based Reference Building and Simulation Dataset. Energy Build. 2022, 258, 111717. [Google Scholar] [CrossRef]
  144. Xu, L.; Hu, M.; Fan, C. Probabilistic Electrical Load Forecasting for Buildings Using Bayesian Deep Neural Networks. J. Build. Eng. 2022, 46, 103853. [Google Scholar] [CrossRef]
  145. Liu, J.; Zhang, Q.; Dong, Z.; Li, X.; Li, G.; Xie, Y.; Li, K. Quantitative Evaluation of the Building Energy Performance Based on Short-Term Energy Predictions. Energy 2021, 223, 120065. [Google Scholar] [CrossRef]
  146. Kim, C.; Kim, M.; Song, Y. Sequence-to-Sequence Deep Learning Model for Building Energy Consumption Prediction with Dynamic Simulation Modeling. J. Build. Eng. 2021, 43, 102577. [Google Scholar] [CrossRef]
  147. Savi, M.; Olivadese, F. Short-Term Energy Consumption Forecasting at the Edge: A Federated Learning Approach. IEEE Access 2021, 9, 95949–95969. [Google Scholar] [CrossRef]
  148. Reddy, S.; Akashdeep, S.; Harshvardhan, R.; Kamath, S. Stacking Deep Learning and Machine Learning Models for Short-Term Energy Consumption Forecasting. Adv. Eng. Inform. 2022, 52, 101542. [Google Scholar]
  149. Zhou, C.; Fang, Z.; Xu, X.; Zhang, X.; Ding, Y.; Jiang, X.; Ji, Y. Using Long Short-Term Memory Networks to Predict Energy Consumption of Air-Conditioning Systems. Sustain. Cities Soc. 2020, 55, 102000. [Google Scholar] [CrossRef]
  150. Mahamedi, E.; Wonders, M.; Seresht, N.; Woo, W.; Kassem, M. A Reinforcing Transfer Learning Approach to Predict Buildings Energy Performance. Constr. Innov. 2023, 24, 242–255. [Google Scholar] [CrossRef]
  151. Durand, D.; Aguilar, J.; R-Moreno, M. An Analysis of the Energy Consumption Forecasting Problem in Smart Buildings Using LSTM. Sustainability 2022, 14, 13358. [Google Scholar] [CrossRef]
  152. Fatehi, N.; Politis, A.; Lin, L.; Stobby, M.; Nazari, M. Machine Learning Based Occupant Behavior Prediction in Smart Building to Improve Energy Efficiency. In Proceedings of the 2023 IEEE Power & Energy Society Innovative Smart Grid Technologies Conference (ISGT), Washington, DC, USA, 16–19 January 2023. [Google Scholar]
  153. Li, G.; Wu, Y.; Liu, J.; Fang, X.; Wang, Z. Performance Evaluation of Short-Term Cross-Building Energy Predictions Using Deep Transfer Learning Strategies. Energy Build. 2022, 275, 112461. [Google Scholar] [CrossRef]
  154. Liu, X.; Li, S.; Fan, Q. Prediction of Building Heating and Cooling Load Based on IPSO-LSTM Neural Network. In Proceedings of the 2020 Chinese Automation Congress (CAC), Shanghai, China, 6–8 November 2020; pp. 1085–1090. [Google Scholar]
  155. Faiq, M.; Tan, K.; Liew, C.; Hossain, F.; Tso, C.; Lim, C.; Wong, A.; Shah, Z. Prediction of Energy Consumption in Campus Buildings Using Long Short-Term Memory. Alex. Eng. J. 2023, 67, 65–76. [Google Scholar] [CrossRef]
  156. Qiao, Q.; Yunusa-Kaltungo, A.; Edwards, R. Preliminary Exploration of Factors Affecting Building Energy Consumption Prediction. In Proceedings of the 2021 IEEE PES/IAS PowerAfrica, Nairobi, Kenya, 23–27 August 2021; pp. 141–145. [Google Scholar]
  157. Yang, Y.; Li, G.; Luo, T.; Al-Bahrani, M.; Al-Ammar, E.; Sillanpaa, M.; Ali, S.; Leng, X. The Innovative Optimization Techniques for Forecasting the Energy Consumption of Buildings Using the Shuffled Frog Leaping Algorithm and Different Neural Networks. Energy 2023, 268, 126548. [Google Scholar] [CrossRef]
  158. Zhou, Y.; Li, X.; Liu, Y.; Wei, R. Transfer Learning-Based Adaptive Recursive Neural Network for Short-Term Non-Stationary Building Heating Load Prediction. J. Build. Eng. 2023, 76, 107271. [Google Scholar] [CrossRef]
  159. Somu, N.; Raman, M.; Ramamritham, K. A Hybrid Model for Building Energy Consumption Forecasting Using Long Short Term Memory Networks. Appl. Energy 2020, 261, 114131. [Google Scholar] [CrossRef]
  160. Jung, S.; Moon, J.; Park, S.; Hwang, E. An Attention-Based Multilayer GRU Model for Multistep-Ahead Short-Term Load Forecasting. Sensors 2021, 21, 1639. [Google Scholar] [CrossRef] [PubMed]
  161. Shan, S.; Cao, B.; Wu, Z. Forecasting the Short-Term Electricity Consumption of Building Using a Novel Ensemble Model. IEEE Access 2019, 7, 88093–88106. [Google Scholar] [CrossRef]
  162. Wen, L.; Zhou, K.; Yang, S. Load Demand Forecasting of Residential Buildings Using a Deep Learning Model. Electr. Power Syst. Res. 2020, 179, 106073. [Google Scholar] [CrossRef]
  163. Liu, H.; Liu, Y.; Guo, X.; Wu, H.; Wang, H.; Liu, Y. An Energy Consumption Prediction Method for HVAC Systems Using Energy Storage Based on Time Series Shifting and Deep Learning. Energy Build. 2023, 298, 113508. [Google Scholar] [CrossRef]
  164. Song, L.; Gao, W.; Yang, Y.; Zhang, L.; Li, Q.; Dong, Z. Terminal Cooling Load Forecasting Model Based on Particle Swarm Optimization. Sustainability 2022, 14, 11924. [Google Scholar] [CrossRef]
  165. Bedi, G.; Venayagamoorthy, G.; Singh, R. Development of an IoT-Driven Building Environment for Prediction of Electric Energy Consumption. IEEE Internet Things J. 2020, 7, 4912–4921. [Google Scholar] [CrossRef]
  166. Skomski, E.; Lee, J.; Kim, W.; Chandan, V.; Katipamula, S.; Hutchinson, B. Sequence-to-Sequence Neural Networks for Short-Term Electrical Load Forecasting in Commercial Office Buildings. Energy Build. 2020, 226, 110350. [Google Scholar] [CrossRef]
  167. Ullah, F.; Ullah, A.; Ul Haq, I.; Rho, S.; Baik, S. Short-Term Prediction of Residential Power Energy Consumption via CNN and Multi-Layer Bi-Directional LSTM Networks. IEEE Access 2020, 8, 123369–123380. [Google Scholar] [CrossRef]
  168. Zhou, J. Convolutional Neural Network Based Energy Consumption Management Model for the Full Life Cycle of Buildings and Information System Design. Mob. Inf. Syst. 2022, 2022, 6181357. [Google Scholar] [CrossRef]
  169. Zhang, C.; Li, J.; Zhao, Y.; Li, T.; Chen, Q.; Zhang, Q. A Hybrid Deep Learning-Based Method for Short-Term Building Energy Load Prediction Combined with an Interpretation Process. Energy Build. 2020, 225, 110301. [Google Scholar] [CrossRef]
  170. Gao, Z.; Yu, J.; Zhao, A.; Hu, Q.; Yang, S. A Hybrid Method of Cooling Load Forecasting for Large Commercial Building Based on Extreme Learning Machine. Energy 2022, 238, 122073. [Google Scholar] [CrossRef]
  171. Fang, X.; Gong, G.; Li, G.; Chun, L.; Li, W.; Peng, P. A Hybrid Deep Transfer Learning Strategy for Short Term Cross-Building Energy Prediction. Energy 2021, 215, 119208. [Google Scholar] [CrossRef]
  172. Zhao, J.; Liu, X. A Hybrid Method of Dynamic Cooling and Heating Load Forecasting for Office Buildings Based on Artificial Intelligence and Regression Analysis. Energy Build. 2018, 174, 293–308. [Google Scholar] [CrossRef]
  173. Karijadi, I.; Chou, S. A Hybrid RF-LSTM Based on CEEMDAN for Improving the Accuracy of Building Energy Consumption Prediction. Energy Build. 2022, 259, 111908. [Google Scholar] [CrossRef]
  174. Li, C.; Tang, M.; Zhang, G.; Wang, R.; Tian, C. A Hybrid Short-Term Building Electrical Load Forecasting Model Combining the Periodic Pattern, Fuzzy System, and Wavelet Transform. Int. J. Fuzzy Syst. 2020, 22, 156–171. [Google Scholar] [CrossRef]
  175. Li, C.; Li, G.; Wang, K.; Han, B. A Multi-Energy Load Forecasting Method Based on Parallel Architecture CNN-GRU and Transfer Learning for Data Deficient Integrated Energy Systems. Energy 2022, 259, 124967. [Google Scholar] [CrossRef]
  176. Huang, Y.; Li, C. Accurate Heating, Ventilation and Air Conditioning System Load Prediction for Residential Buildings Using Improved Ant Colony Optimization and Wavelet Neural Network. J. Build. Eng. 2021, 35, 101972. [Google Scholar] [CrossRef]
  177. Yuan, Z.; Wang, W.; Wang, H.; Mizzi, S. Combination of Cuckoo Search and Wavelet Neural Network for Midterm Building Energy Forecast. Energy 2020, 202, 117728. [Google Scholar] [CrossRef]
  178. Shao, X.; Pu, C.; Zhang, Y.; Kim, C. Domain Fusion CNN-LSTM for Short-Term Power Consumption Forecasting. IEEE Access 2020, 8, 188352–188362. [Google Scholar] [CrossRef]
  179. Syed, D.; Abu-Rub, H.; Ghrayeb, A.; Refaat, S. Household-Level Energy Forecasting in Smart Buildings Using a Novel Hybrid Deep Learning Model. IEEE Access 2021, 9, 33498–33511. [Google Scholar] [CrossRef]
  180. Tian, Y.; Yu, J.; Zhao, A. Predictive Model of Energy Consumption for Office Building by Using Improved GWO-BP. Energy Rep. 2020, 6, 620–627. [Google Scholar] [CrossRef]
  181. Hu, Y.; Cheng, X.; Wang, S.; Chen, J.; Zhao, T.; Dai, E. Times Series Forecasting for Urban Building Energy Consumption Based on Graph Convolutional Network. Appl. Energy 2022, 307, 118231. [Google Scholar] [CrossRef]
  182. Khan, Z.; Hussain, T.; Ullah, A.; Rho, S.; Lee, M.; Baik, S. Towards Efficient Electricity Forecasting in Residential and Commercial Buildings: A Novel Hybrid CNN with a LSTM-AE Based Framework. Sensors 2020, 20, 1399. [Google Scholar] [CrossRef]
  183. Liu, Z.; Yu, J.; Feng, C.; Su, Y.; Dai, J.; Chen, Y. A Hybrid Forecasting Method for Cooling Load in Large Public Buildings Based on Improved Long Short Term Memory. J. Build. Eng. 2023, 76, 107238. [Google Scholar] [CrossRef]
  184. Ruan, Y.; Wang, G.; Meng, H.; Qian, F. A Hybrid Model for Power Consumption Forecasting Using VMD-Based the Long Short-Term Memory Neural Network. Front. Energy Res. 2022, 9, 772508. [Google Scholar] [CrossRef]
  185. Mao, Y.; Yu, J.; Zhang, N.; Dong, F.; Wang, M.; Li, X. A Hybrid Model of Commercial Building Cooling Load Prediction Based on the Improved NCHHO-FENN Algorithm. J. Build. Eng. 2023, 78, 107660. [Google Scholar] [CrossRef]
  186. Duan, Y. A Novel Interval Energy-Forecasting Method for Sustainable Building Management Based on Deep Learning. Sustainability 2022, 14, 8584. [Google Scholar] [CrossRef]
  187. Wang, L.; Xie, D.; Zhou, L.; Zhang, Z. Application of the Hybrid Neural Network Model for Energy Consumption Prediction of Office Buildings. J. Build. Eng. 2023, 72, 106503. [Google Scholar] [CrossRef]
  188. Irshad, K.; Zahir, M.; Shaik, M.; Ali, A. Buildings’ Heating and Cooling Load Prediction for Hot Arid Climates: A Novel Intelligent Data-Driven Approach. Buildings 2022, 12, 1677. [Google Scholar] [CrossRef]
  189. Gaur, K.; Singh, S. CNN-Bi-LSTM Based Household Energy Consumption Prediction. In Proceedings of the 2021 3rd International Conference on Signal Processing and Communication (ICPSC), Coimbatore, India, 13–14 May 2021; pp. 233–237. [Google Scholar]
  190. Abdelaziz, A.; Santos, V.; Dias, M. Convolutional Neural Network with Genetic Algorithm for Predicting Energy Consumption in Public Buildings. IEEE Access 2023, 11, 64049–64069. [Google Scholar] [CrossRef]
  191. Ntsaluba, S.; Nwulu, N.; Ntsaluba, K. Deep Learning Based Building Load Prediction for Residences of an Academic Institutions. In Proceedings of the 2021 International Conference on Electrical, Computer and Energy Technologies (ICECET), Cape Town, South Africa, 9–10 December 2021; pp. 2284–2291. [Google Scholar]
  192. Palak, M.; Revati, G.; Hossain, M.A.; Sheikh, A. Deep Learning Models for Smart Building Load Profile Prediction. In Proceedings of the 2021 31st Australasian Universities Power Engineering Conference (AUPEC), Perth, Australia, 26–30 September 2021; pp. 1–6. [Google Scholar]
  193. Tomar, D.; Tomar, P.; Bhardwaj, A.; Sinha, G. Deep Learning Neural Network Prediction System Enhanced with Best Window Size in Sliding Window Algorithm for Predicting Domestic Power Consumption in a Residential Building. Comput. Intell. Neurosci. 2022, 2022, 7216959. [Google Scholar] [CrossRef]
  194. Guo, J.; Lin, P.; Zhang, L.; Pan, Y.; Xiao, Z. Dynamic Adaptive Encoder-Decoder Deep Learning Networks for Multivariate Time Series Forecasting of Building Energy Consumption. Appl. Energy 2023, 350, 121803. [Google Scholar] [CrossRef]
  195. Barzola-Monteses, J.; Guerrero, M.; Parrales-Bravo, F.; Espinoza-Andaluz, M. Forecasting Energy Consumption in Residential Department Using Convolutional Neural Networks. In Proceedings of the Conference on Information and Communication Technologies of Ecuador, Guayaquil, Ecuador, 24–26 November 2021; Volume 1456, pp. 18–30. [Google Scholar]
  196. An, W.; Zhu, X.; Yang, K.; Kim, M.; Liu, J. Hourly Heat Load Prediction for Residential Buildings Based on Multiple Combination Models: A Comparative Study. Buildings 2023, 13, 2340. [Google Scholar] [CrossRef]
  197. Sretenovic, A.; Jovanovic, R.; Novakovic, V.; Nord, N. Hybrid artificial intelligence model for prediction of heating energy use. Therm. Sci. 2022, 26, 705–716. [Google Scholar] [CrossRef]
  198. Kavitha, R.; Thiagarajan, C.; Priya, P.; Anand, A.; Al-Ammar, E.; Santhamoorthy, M.; Chandramohan, P. Improved Harris Hawks Optimization with Hybrid Deep Learning Based Heating and Cooling Load Prediction on Residential Buildings. Chemosphere 2022, 309, 136525. [Google Scholar] [CrossRef]
  199. Wang, H.; Ma, W.; Wang, Z.; Lu, C. Multiscale Convolutional Recurrent Neural Network for Residential Building Electricity Consumption Prediction. J. Intell. Fuzzy Syst. 2022, 43, 3479–3491. [Google Scholar] [CrossRef]
  200. Chen, P.; Chen, L. Prediction Method of Intelligent Building Electricity Consumption Based on Deep Learning. Evol. Intell. 2023, 16, 1637–1644. [Google Scholar] [CrossRef]
  201. Dong, F.; Yu, J.; Quan, W.; Xiang, Y.; Li, X.; Sun, F. Short-Term Building Cooling Load Prediction Model Based on DwdAdam-ILSTM Algorithm: A Case Study of a Commercial Building. Energy Build. 2022, 272, 112337. [Google Scholar] [CrossRef]
  202. Alymani, M.; Mengash, H.; Aljebreen, M.; Alasmari, N.; Allafi, R.; Alshahrani, H.; Elfaki, M.; Hamza, M.; Abdelmageed, A. Sustainable Residential Building Energy Consumption Forecasting for Smart Cities Using Optimal Weighted Voting Ensemble Learning. Sustain. Energy Technol. Assess. 2023, 57, 103271. [Google Scholar] [CrossRef]
  203. Kim, J.; Moon, J.; Hwang, E.; Kang, P. Recurrent Inception Convolution Neural Network for Multi Short-Term Load Forecasting. Energy Build. 2019, 194, 328–341. [Google Scholar] [CrossRef]
  204. Ngo, N.; Pham, A.; Truong, T.; Truong, N.; Huynh, N.; Pham, T. An Ensemble Machine Learning Model for Enhancing the Prediction Accuracy of Energy Consumption in Buildings. Arab. J. Sci. Eng. 2022, 47, 4105–4117. [Google Scholar] [CrossRef]
  205. Moon, J.; Jung, S.; Rew, J.; Rho, S.; Hwang, E. Combination of Short-Term Load Forecasting Models Based on a Stacking Ensemble Approach. Energy Build. 2020, 216, 109921. [Google Scholar] [CrossRef]
  206. Lee, J.; Kim, J.; Ko, W. Day-Ahead Electric Load Forecasting for the Residential Building with a Small-Size Dataset Based on a Self-Organizing Map and a Stacking Ensemble Learning Method. Appl. Sci. 2019, 9, 1231. [Google Scholar] [CrossRef]
  207. Khan, Z.; Ullah, A.; Ullah, W.; Rho, S.; Lee, M.; Baik, S. Electrical Energy Prediction in Residential Buildings for Short-Term Horizons Using Hybrid Deep Learning Strategy. Appl. Sci. 2020, 10, 8634. [Google Scholar] [CrossRef]
  208. Sekhar Roy, S.; Roy, R.; Balas, V.E. Estimating Heating Load in Buildings Using Multivariate Adaptive Regression Splines, Extreme Learning Machine, a Hybrid Model of MARS and ELM. Renew. Sustain. Energy Rev. 2018, 82, 4256–4268. [Google Scholar] [CrossRef]
  209. Alobaidi, M.; Chebana, F.; Meguid, M. Robust Ensemble Learning Framework for Day-Ahead Forecasting of Household Based Energy Consumption. Appl. Energy 2018, 212, 997–1012. [Google Scholar] [CrossRef]
  210. Ishaq, M.; Kwon, S. Short-Term Energy Forecasting Framework Using an Ensemble Deep Learning Approach. IEEE Access 2021, 9, 94262–94271. [Google Scholar]
  211. Ves, A.V.; Ghitescu, N.; Pop, C.; Antal, M.; Cioara, T.; Anghel, L.; Salomie, L. A Stacking Multi-Learning Ensemble Model for Predicting Near Real Time Energy Consumption Demand of Residential Buildings. In Proceedings of the IEEE 15th International Conference on Intelligent Computer Communication and Processing (ICCP), Cluj-Napoca, Romania, 5–7 September 2019; pp. 183–189. [Google Scholar]
  212. Adnan, W.; Dahlan, N.; Musirin, I. Artificial Neural Network Optimization in Prediction Baseline Energy Consumption to Quantify Energy Savings in Commercial Building. In Proceedings of the 2020 11th IEEE Control and System Graduate Research Colloquium (ICSGRC), Shah Alam, Malaysia, 8 August 2020; pp. 393–397. [Google Scholar]
  213. Yu, J.; Zhang, T.; Zhao, A.; Xie, Y. Research on Energy Consumption Prediction of Office Buildings Based on Comprehensive Similar Day and Ensemble Learning. J. Intell. Fuzzy Syst. 2021, 40, 11951–11965. [Google Scholar] [CrossRef]
  214. Cao, W.; Yu, J.; Chao, M.; Wang, J.; Yang, S.; Zhou, M.; Wang, M. Short-Term Energy Consumption Prediction Method for Educational Buildings Based on Model Integration. Energy 2023, 283, 128580. [Google Scholar] [CrossRef]
  215. Wang, R.; Lu, S.; Feng, W. A Novel Improved Model for Building Energy Consumption Prediction Based on Model Integration. Appl. Energy 2020, 262, 114561. [Google Scholar] [CrossRef]
  216. Sharif, S.; Hammad, A. Developing Surrogate ANN for Selecting Near-Optimal Building Energy Renovation Methods Considering Energy Consumption, LCC and LCA. J. Build. Eng. 2019, 25, 100790. [Google Scholar] [CrossRef]
  217. Moon, J.; Jung, S. Development of a Thermal Control Algorithm Using Artificial Neural Network Models for Improved Thermal Comfort and Energy Efficiency in Accommodation Buildings. Appl. Therm. Eng. 2016, 103, 1135–1144. [Google Scholar] [CrossRef]
  218. Bagheri-Esfeh, H.; Safikhani, H.; Motahar, S. Multi-Objective Optimization of Cooling and Heating Loads in Residential Buildings Integrated with Phase Change Materials Using the Artificial Neural Network and Genetic Algorithm. J. Energy Storage 2020, 32, 101772. [Google Scholar] [CrossRef]
  219. Zekic-Susac, M.; Has, A.; Knezevic, M. Predicting Energy Cost of Public Buildings by Artificial Neural Networks, CART, and Random Forest. Neurocomputing 2021, 439, 223–233. [Google Scholar] [CrossRef]
  220. Xu, Y.; Li, F.; Asgari, A. Prediction and Optimization of Heating and Cooling Loads in a Residential Building Based on Multi-Layer Perceptron Neural Network and Different Optimization Algorithms. Energy 2022, 240, 122692. [Google Scholar] [CrossRef]
  221. Sui, Z.; Mu, J.; Wang, T.; Zhang, S. Evaluation of Energy Saving of Residential Buildings in North China Using Back-Propagation Neural Network and Virtual Reality Modeling. J. Energy Eng. 2022, 148, 04022013. [Google Scholar] [CrossRef]
  222. Yang, Y.; Liu, X.; Tian, C. Optimization Method for Energy Saving of Rural Architectures in Hot Summer and Cold Winter Areas Based on Artificial Neural Network. Comput. Intell. Neurosci. 2022, 2022, 2232425. [Google Scholar] [CrossRef] [PubMed]
  223. Wan, T.; Bai, Y.; Wang, T.; Wei, Z. BPNN-Based Optimal Strategy for Dynamic Energy Optimization with Providing Proper Thermal Comfort under the Different Outdoor Air Temperatures. Appl. Energy 2022, 313, 118899. [Google Scholar] [CrossRef]
  224. Yang, S.; Wan, M.; Chen, W.; Ng, B.; Dubey, S. Model Predictive Control with Adaptive Machine-Learning-Based Model for Building Energy Efficiency and Comfort Optimization. Appl. Energy 2020, 271, 115147. [Google Scholar] [CrossRef]
  225. Ilbeigi, M.; Ghomeishi, M.; Dehghanbanadaki, A. Prediction and Optimization of Energy Consumption in an Office Building Using Artificial Neural Network and a Genetic Algorithm. Sustain. Cities Soc. 2020, 61, 102325. [Google Scholar] [CrossRef]
  226. Garcia Kerdan, I.; Morillon Galvez, D. Artificial Neural Network Structure Optimisation for Accurately Prediction of Exergy, Comfort and Life Cycle Cost Performance of a Low Energy Building. Appl. Energy 2020, 280, 115862. [Google Scholar] [CrossRef]
  227. Seyedzadeh, S.; Rahimian, F.; Oliver, S.; Glesk, I.; Kumar, B. Data Driven Model Improved by Multi-Objective Optimisation for Prediction of Building Energy Loads. Autom. Constr. 2020, 116, 103188. [Google Scholar] [CrossRef]
  228. Li, L.; Fu, Y.; Fung, J.; Lau, A. Development of a Back-Propagation Neural Network and Adaptive Grey Wolf Optimizer Algorithm for Thermal Comfort and Energy Consumption Prediction and Optimization. Energy Build. 2021, 253, 111439. [Google Scholar] [CrossRef]
  229. Lin, Y.; Zhou, S.; Yang, W.; Shi, L.; Li, C. Development of Building Thermal Load and Discomfort Degree Hour Prediction Models Using Data Mining Approaches. Energies 2018, 11, 1570. [Google Scholar] [CrossRef]
  230. Hosamo, H.; Hosamo, M.; Nielsen, H.; Svennevig, P.; Svidt, K. Digital Twin of HVAC System (HVACDT) for Multiobjective Optimization of Energy Consumption and Thermal Comfort Based on BIM Framework with ANN-MOGA. Adv. Build. Energy Res. 2023, 17, 125–171. [Google Scholar] [CrossRef]
  231. Zhan, J.; He, W.; Huang, J. Dual-Objective Building Retrofit Optimization under Competing Priorities Using Artificial Neural Network. J. Build. Eng. 2023, 70, 106376. [Google Scholar] [CrossRef]
  232. Wang, N. Multi-Objective Energy Saving Optimization of Residential Buildings Based on MABC-BP. Energy Rep. 2023, 9, 922–929. [Google Scholar] [CrossRef]
  233. Zhang, J.; Ji, L. Optimization and Prediction of Energy Consumption, Daylighting, and Thermal Comfort of Buildings in Tropical Areas. Adv. Civ. Eng. 2022, 2022, 3178269. [Google Scholar] [CrossRef]
  234. Zhao, Y.; Qian, S.; Jing, C. Prediction and Optimization of Indoor Thermal Environment and Energy Consumption Based on Artificial Neural Network. In IOP Conference Series: Earth and Environmental Science; IOP Publishing: Bristol, UK, 2020; Volume 463. [Google Scholar]
  235. Li, K.; Zhang, J.; Chen, X.; Xue, W. Building’s Hourly Electrical Load Prediction Based on Data Clustering and Ensemble Learning Strategy. Energy Build. 2022, 261, 111943. [Google Scholar] [CrossRef]
  236. Ahmad, T.; Chen, H.; Shair, J.; Chengliang, X. Deployment of Data-Mining Short and Medium-Term Horizon Cooling Load Forecasting Models for Building Energy Optimization and Management. Int. J. Refrig. Rev. Int. Du Froid 2019, 98, 399–409. [Google Scholar] [CrossRef]
  237. Szul, T.; Tabor, S.; Pancerz, K. Application of the BORUTA Algorithm to Input Data Selection for a Model Based on Rough Set Theory (RST) to Prediction Energy Consumption for Building Heating. Energies 2021, 14, 2779. [Google Scholar] [CrossRef]
  238. Tian, Y.; Sehovac, L.; Grolinger, K. Similarity-Based Chained Transfer Learning for Energy Forecasting with Big Data. IEEE Access 2019, 7, 139895–139908. [Google Scholar] [CrossRef]
  239. Luo, X.; Oyedele, L.; Ajayi, A.; Monyei, C.; Akinade, O.; Akanbi, L. Development of an IoT-Based Big Data Platform for Day-Ahead Prediction of Building Heating and Cooling Demands. Adv. Eng. Inform. 2019, 41, 100926. [Google Scholar] [CrossRef]
  240. Sholahudin, S.; Han, H. Simplified Dynamic Neural Network Model to Predict Heating Load of a Building Using Taguchi Method. Energy 2016, 115, 1672–1678. [Google Scholar] [CrossRef]
  241. Cho, H.; Nam, Y.; Choi, E.; Choi, Y.; Kim, H.; Bae, S.; Moon, S. Comparative Analysis of the Optimized ANN, SVM, and Tree Ensemble Models Using Bayesian Optimization for Predicting GSHP COP. J. Build. Eng. 2021, 44, 103411. [Google Scholar] [CrossRef]
  242. Cheng, Z.; Wang, L.; Cheng, Y.; Chen, G. Heat Prediction of High Energy Physical Data Based on LSTM Recurrent Neural Network. In Proceedings of the 24th International Conference on Computing in High Energy and Nuclear Physics (CHEP 2019), Adelaide, Australia, 4–8 November 2019; Volume 245. [Google Scholar]
  243. Wu, G.; Zhang, L.; Lin, T.; Hao, C. Forecast of Total Rural Energy Demand Based on Energy Consumption Intensity. Front. Energy Res. 2023, 10, 1021164. [Google Scholar]
  244. Zhou, Y.; Wang, L.; Qian, J. Application of Combined Models Based on Empirical Mode Decomposition, Deep Learning, and Autoregressive Integrated Moving Average Model for Short-Term Heating Load Predictions. Sustainability 2022, 14, 7349. [Google Scholar] [CrossRef]
Sustainability 16 07805 g001
Figure 1. The total number of papers published.
Figure 1. The total number of papers published.
Sustainability 16 07805 g001
Sustainability 16 07805 g002
Figure 2. Literature classification.
Figure 2. Literature classification.
Sustainability 16 07805 g002
Sustainability 16 07805 g003
Figure 3. Flowchart.
Figure 3. Flowchart.
Sustainability 16 07805 g003
Sustainability 16 07805 g004
Figure 4. Number of papers published about different ANNs.
Figure 4. Number of papers published about different ANNs.
Sustainability 16 07805 g004
Sustainability 16 07805 g005
Figure 5. Application distribution of neural network energy consumption prediction model in different building stages.
Figure 5. Application distribution of neural network energy consumption prediction model in different building stages.
Sustainability 16 07805 g005
Sustainability 16 07805 g006
Figure 6. Application in the early design stage of the building: (a) building type, (b) energy consumption type, (c) evaluation index, (d) data type.
Figure 6. Application in the early design stage of the building: (a) building type, (b) energy consumption type, (c) evaluation index, (d) data type.
Sustainability 16 07805 g006
Sustainability 16 07805 g007
Figure 7. The number of papers published concerning the early design stage of the building based on different ANNs.
Figure 7. The number of papers published concerning the early design stage of the building based on different ANNs.
Sustainability 16 07805 g007
Sustainability 16 07805 g008
Figure 8. Application of building operation stage: (a) time span, (b) building type, (c) energy consumption type, (d) evaluation index.
Figure 8. Application of building operation stage: (a) time span, (b) building type, (c) energy consumption type, (d) evaluation index.
Sustainability 16 07805 g008
Sustainability 16 07805 g009
Figure 9. The number of papers published concerning the building operation stage based on different ANNs.
Figure 9. The number of papers published concerning the building operation stage based on different ANNs.
Sustainability 16 07805 g009
Sustainability 16 07805 g010
Figure 10. Application in the building renovation stage: (a) building type, (b) energy consumption type, (c) evaluation index, (d) data type.
Figure 10. Application in the building renovation stage: (a) building type, (b) energy consumption type, (c) evaluation index, (d) data type.
Sustainability 16 07805 g010
Sustainability 16 07805 g011
Figure 11. Flow diagram of building energy consumption prediction based on artificial neural network.
Figure 11. Flow diagram of building energy consumption prediction based on artificial neural network.
Sustainability 16 07805 g011
Table 1. Characteristics of common neural networks under FFNN structure.
Table 1. Characteristics of common neural networks under FFNN structure.
StructurePeculiarityMeritShortcoming
MLPFully interconnected neurons between adjacent layersCapable of learning complex nonlinear relationships;
Applicable to diverse machine learning tasks.		Highly flexible in problem-solving;
Efficient training using backpropagation.		Prolonged training times for large datasets and complex architectures;
Manual configuration of network structure and hyperparameters required.
WNNIntegration of wavelet transform with neural networksUtilizes wavelet transform for feature extraction and neural networks for learning and classification.Extracts local features and frequency information from data;
Exhibits robust generalization capabilities.		Manual selection of wavelet basis function and network configuration necessary;
Extended training durations.
ELMRandom initialization of weights in the hidden layerRandom initialization of hidden layer weights, analytical determination of output layer weights.Rapid training with no iterative hidden layer weight adjustments;
Less likely to converge to local optima in complex scenarios.		Potential for overfitting training data;
Limited generalization capability in complex problem settings.
NARXDynamic system utilizing past network outputs and external inputsSuitable for time series forecasting, capturing nonlinear temporal relationships.Efficiently processes time series data;
Applicable to various time series forecasting tasks.		Manual selection of model structures and parameters is essential;
May lack efficiency for lengthy and high-dimensional time series.
Table 2. Characteristics of common neural networks under RNN structure.
Table 2. Characteristics of common neural networks under RNN structure.
StructurePeculiarityMeritShortcoming
LSTMInput gate, forget gate, output gate, and memory cellAddresses gradient vanishing and explosion in traditional RNNs, controlling information flow.Capable of processing long sequence data and capturing long-term dependencies;
Possesses strong memory capabilities.		Numerous parameters, relatively complex model, and extended training time;
Manual adjustment of network structure and hyperparameters required.
GRUUpdate gate and reset gateReduces parameter count, enhances training efficiency, and controls information flow.Faster training, comparable to LSTM in some cases but with lower computational cost.May lack flexibility for certain complex sequence data, with limited capturing capability.
ElmanInput layer, hidden layer, and output layerSimple recurrent connections, suitable for basic sequence data modeling tasks.Simple structure, easy to understand and implement;
Performs well in simple sequence modeling tasks.		Prone to gradient vanishing or exploding issues;
Less effective than LSTM and GRU in handling complex sequence data.
Bidirectional RNNBidirectional hidden layers, implemented via forward and backward recurrent connectionsCombines forward and backward information flows to better capture bidirectional dependencies in sequence data, allowing for a comprehensive consideration of both past and future information.Effectively captures bidirectional dependencies within sequence data;
Excels in certain sequence modeling tasks.		Higher computational complexity and model complexity;
Requires more data and training time.
Table 3. Application of ANN-based energy consumption prediction in the early design stage of buildings.
Table 3. Application of ANN-based energy consumption prediction in the early design stage of buildings.
Ref.Type of BuildingVariableType of Energy ConsumptionMethod
Architectural design[21]Office buildingsLength, width, height, overhang length, window-to-wall ratio, orientation, wall u-value, window u-value, ground floor u-value, roof u-value, window g-value, floor heat capacity, infiltration air change rate, number of floors, lighting heat gain, equipment heat gain, chiller coefficient of performance, boiler efficiency, chiller type, boiler pump type.Energy consumptionCNN
[22]Residential buildingsTemperature, wind speed, pressure, postcode, total floor area, property type, glazed area, extension count, wall description, floor description, floor level, window description, window energy efficiency, window environmental efficiency, wall energy efficiency, wall environmental efficiency, roof description, roof energy efficiency, roof environmental efficiency, lighting environmental efficiency, lighting energy efficiency, number of heated rooms, number of habitable rooms, energy rating.Energy consumptionSVM, GB, RF, DT, KNN, ET, GP, MLP, AdaBoost
[23]Residential buildingsLength, depth, height, building orientation, window-to-wall ratio, glazing type, temperature setpoint.Energy consumption and electricity consumption for cooling and heatingMLP-LM
[24]Office buildingsFloor area, floor height, no. of floors, orientation, u-value (wall), u-value (ground floor), u-value (roof), u-value (window), g-value, window-to-wall ratio, internal mass, air permeability, occupant load, light heat load, heating COP, cooling COP, boiler efficiency.Energy consumptionCNN
[25]Medical buildingsExternal plaster, the thickness of the external plaster, insulation material, the thickness of the insulation material, wall material, wall material thickness, internal plaster, internal plaster thickness, window type, pilot region.Energy consumptionGA-ANN
[26]Standard unitBuilding type, floor surface area, glazed zone, floor level, number of habitable rooms, number of air-conditioned rooms, ceiling height, average monthly temperature, average monthly pressure, average monthly humidity, average monthly wind speed.Electricity consumptionANN, SVM, KNN, DT, RF, GB, LR, DNN
[27]Standard unitLength, width, floor height, window-to-wall ratio, number of floors, number of rooms.Energy consumption for cooling, heating, and lightingMLP
[28]Office buildingsOrientation, building facade, window-to-wall ratio, area of each floor, floor height, number of floors, envelope density, conductivity, specific heat, thickness, glass u-factor, glass SHGC. Heating and cooling loadsMLP, GA-NN
[29]Rural architectureWidth, floor height, depth, building orientation, building window–wall ratio.Energy consumptionMLP, SVR
[30]Campus buildingsRoom depth, standard floor height, width of corridor, orientation, WWR of north external wall, WWR of south external wall.Energy consumption for lighting and heatingBPNN
Early prediction[31]Residential buildingsWall, roof, floor dimensions, opening sizes and orientations, construction type and levels of insulation, weather conditions, location, ventilation and air tightness features, heating system supply. Energy consumptionFFNN
[32]Residential buildingsHeight, relative compactness, wall surface, building surface, orientation, window-to-wall ratio.Heating and cooling loadsMLP-GD
[33]Residential buildingsGlazing area distribution, relative compactness, overall height, surface area, roof area, wall area, orientation, glazing area.Heating and cooling loadsMLP-GOA, MLP-GWO
[34]Residential buildingsIndoor average temperature, average air temperature, radiant temperature, relative humidity, relative compactness, outdoor dry-bulb temperature, rainfall and wind speed.Energy consumption for heatingMLP-LM
[35]Residential buildingsExterior wall heat transfer coefficient, roof heat transfer coefficient, ground floor heat transfer coefficient, window heat transfer coefficient, solar heat gain coefficient (SHGC), WWR on east facades, WWR on west facades, WWR on south facades, WWR on north facades, indoor lighting, electrical equipment, personnel activities.Electricity consumptionBP
[36]Residential buildingsRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distribution.Heating and cooling loadsMLP, SVR
[37]Residential buildingsRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distribution.Cooling loadSFS-MLP
[38]Residential buildingsRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distribution.Energy consumption for heatingBP-MLP, VSA-MLP, BSA-MLP
[39]regionThe building shape factor (BSF), window–wall ratio (WWR), indoor design temperatures in summer and winter, occupancy density, lighting power density, equipment power density.Heating and cooling loads1D-CNN
[40]Residential buildingsExternal walls’ coefficient of transmission, roof coefficient of transmission, floor coefficient of transmission, external wall coefficient of solar radiation absorption, roof coefficient of solar radiation absorption, thermal bridge linear coefficient, rate of air change, window-to-wall ratio for all directions (i.e., north, east, west, and south), glazing.Energy consumption for heatingMTOA-MLP
[41]Commercial buildingsWeather variable and daysRefrigeration, lighting energy consumption, and total electricity consumptionMLP
[42]Residential buildingsShape factor, volume, total window area, opaque surface area, cooling degree day, heating degree days, relative humidity, wind speed, and solar radiation, PCM melting point. Energy consumption of PCM-integrated buildingsMR, SVM, MLP
[43]Standard unitRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, Gglazing area distributionHeating and cooling loadsBayesian
[44]Residential buildingsRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, Gglazing area distributionHeating and cooling loadsLR, SVR, MLP, DGAM
[45]Many types of buildingsRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distributionEnergy consumption for heatingABC-ANN, PSO-ANN, ICA-ANN, GA-ANN
[46]Commercial buildingsRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distribution.Heating and cooling loadsEFA-MLP
[47]Office buildingsOutdoor air dry-bulb temperature; outdoor air humidity ratio; global horizontal radiation–cloud cover ratio; occupancy schedule for zones 1A and 1B; occupancy schedule for zone 2; office equipment schedule; indoor air dry-bulb temperature; BIPV temperature: last time step; BIPV temperature: same time step, previous day; heating load: last time step; heating load: time steps i-2, i-3, i-4; heating load: same time step, previous day; cooling load: last time step; cooling load: time steps i-2, i-3, i-4; cooling load: same time step, previous day; lighting load: last time step.Heating and cooling loads, lighting energy consumption, and BIPV power generationMLP, SVM, LSTM
Table 4. Application of ANN-based energy consumption prediction in the building operation stage.
Table 4. Application of ANN-based energy consumption prediction in the building operation stage.
TypeRef.Percentage
FFNNMLP[1,36,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118] 37%
WNN[119,120]1%
ELM[121,122]1%
NARX[113,122,123,124,125]2.5%
RBF[48,126,127]1.5%
RNNLSTM[15,59,88,89,95,99,125,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159]20%
GRU[132,133,142,144,152,158,160,161,162,163,164]5.5%
Elman[48,98,165]1.5%
Recurrent neural networks[102,135,144,166]2%
CNNCNN[18,20,88,99,102,134,167,168]4%
Hybrid modelHybrid model[19,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203]18%
Ensemble learningEnsemble learning[204,205,206,207,208,209,210,211,212,213,214,215]6%
Table 5. Application of ANN-based energy consumption prediction in the building renovation stage.
Table 5. Application of ANN-based energy consumption prediction in the building renovation stage.
Ref.Type of BuildingVariableOptimization TargetsMethod
[216]Campus buildingsRoof surfaces, exterior walls, windows, airtightness, operation schedule, space allocation, activity, HVAC system, temperature setting, heating, cooling, DHW.Total energy consumption (TEC), LCC, and Llifec Cycle Assessment assessment (LCA)SBMO-MLP
[217]Residential buildingsAverage indoor temperature, uncomfortable minutes after normal setpoint period began, total operating minutes of the cooling system, amount of energy.Thermal comfort and energy consumptionMLP
[218]Residential buildingsMelting temperature and thickness of PCM, thermal resistance of exterior walls, internal gain, infiltration rate.Heating and cooling loadsGMDH
[219]Residential buildingsHeated surface of the building, heated volume of the building, flat gross floor area of building, number of employees, number of users, number of floors, factor of building shape F0, total installed internal lighting power, total number of internal lighting lamps.Heating and cooling loadsLM-BPNN
[220]Residential buildingsRelative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distribution.Heating and cooling loadsBBO-MLP, GA-MLP, PSO-MLP, PBIL-MLP, ES-MLP, ACO-MLP
[221]Residential buildingsEnvelope structure.Energy consumptionBPNN
[222]Rural architectureFeatures, layers, height, aboveground area, compass direction, factor, roof, external wall.Energy consumptionBPNN
[223]Office buildingsSupply vane angle, supply air temperature, outdoor air temperature, clothing insulation.Energy consumption, PMV, and DRBPNN-RSM
[224]Office buildingsCooling power, operation status of ACMV and lighting, outdoor air temperature, outdoor RH, estimated number of indoor occupants.Electricity consumptionNARX
[225]Office buildingsIteration number, wall u-value, equipment load rate, infiltration rate, lighting density, number of occupants, roof u-value.Energy consumptionMLP-LM, MLP-GA
[226]Residential buildingsWeather data, HVAC system, cooling setpoint, building orientation, solar protection, glazing, wall insulation/roof insulation, air gap, lighting, solar photovoltaic, solar thermal, wind energy.Thermal comfort, cost, and energy consumptionMLP, GA-NN
[227]Residential and commercial buildingsAverage u-value of envelope, ratio of window area to wall area, form factor, roof ratio, annual sum of energy gained due to infiltration, annual sum of internal heat gain, annual sum of heating degree days, median dry-bulb temperature, annual average of dry point temperature, interquartile range of dew point temperature, annual sum of global horizontal irradiation, annual average of direct normal irradiation, interquartile range of direct normal irradiation.Heating and cooling loadsMOO, RF
[228]Residential buildingsInlet velocity, inlet type, supply air temperature, outlet type, wall heat transfer coefficient, wall thickness, heat flux of heat sources, outdoor temperature.Indoor average temperature and PMVBPNN-GWO
[229]Residential buildingsWindow-to-wall ratio (WWR), absorptance of solar radiation, insulation thickness, concrete thickness.Annual heat load and indoor comfortMLR, CHAID, ECHAID, BPNN, RBFN, CART, SVM
[230]Office buildingsSupply air temperature for each thermal zone (Tui), supply air static pressure (SAP), supply airflow rate (Mai), supply water flow rate in summer (Mcw), supply water flow rate in winter (Mhw), return air temperature (Tuo), supply cooling water temperature (Twi), return cooling water temperature (Two), supply heating water temperature (Thi), return heating water temperature (Tho), ambient temperature (Tai), temperature after rotary heat exchanger (Ti) sensors.Energy consumption and thermal comfortMLP
[231]Campus buildingsWall u-value, wall solar absorptivity, window u-value, window solar heat gain coefficient, window visible transmittance, window-to-wall ratio of all directions (i.e., north, east, west, and south), roof u-value, roof solar absorptivity, ground slab u-value, cooling and heating setpoint temperatures, having been selected in accordance with the prevailing literature, weather data, construction material properties, HVAC properties, occupant activities.Energy consumption and thermal comfortMLP
[232]Residential buildingsWindow-to-wall ratio for all directions (i.e., north, east, west, and south), width of sunroom, thickness of exterior wall insulation layer, thickness of roof insulation layer.Energy consumptionMABC-BP
[233]Office buildingsWindow-to-wall ratio of all directions (i.e., north, east, west, and south), window height, shading device size.Daylighting, energy consumption, and thermal comfortBPNN
[234]Residential buildingsScenarios of different insulation thickness, shading coefficient, ventilation rateEnergy consumption and thermal comfortANN
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Yin, Q.; Han, C.; Li, A.; Liu, X.; Liu, Y. A Review of Research on Building Energy Consumption Prediction Models Based on Artificial Neural Networks. Sustainability 2024, 16, 7805. https://doi.org/10.3390/su16177805

AMA Style

Yin Q, Han C, Li A, Liu X, Liu Y. A Review of Research on Building Energy Consumption Prediction Models Based on Artificial Neural Networks. Sustainability. 2024; 16(17):7805. https://doi.org/10.3390/su16177805

Chicago/Turabian Style

Yin, Qing, Chunmiao Han, Ailin Li, Xiao Liu, and Ying Liu. 2024. "A Review of Research on Building Energy Consumption Prediction Models Based on Artificial Neural Networks" Sustainability 16, no. 17: 7805. https://doi.org/10.3390/su16177805

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop