Search Results (7,783)

A learning management system (LMS) plays a crucial role in supporting students’ educational activities by centralized platforms for course delivery, communication, and student support. Recently, many universities have integrated chatbots into their LMS to assist students with various inquiries and tasks. However, existing chatbots often necessitate human interventions to manually respond to complex queries, resulting in limited scalability and efficiency. In this paper, we present a memory-augmented large language model (LLM) framework that enhances the reasoning and contextual continuity of LMS-based chatbots. The proposed framework first embeds user queries and retrieves semantically relevant entries from various LMS resources, including instructional documents and academic frequently asked questions. Retrieved entries are then filtered through a two-stage confidence filtering process that combines similarity thresholds and LLM-based semantic validation. Validated information, along with user queries, is processed by LLM for response generation. To maintain coherence in multi-turn interactions, the chatbot incorporates short-term, long-term, and temporal event memories, which track conversational flow and personalize responses based on user-specific information, such as recent activity history and individual preferences. To evaluate response quality, we employed a multi-layered evaluation strategy combining BERTScore-based quantitative measurement, an LLM-as-a-Judge approach for automated semantic assessment, and a user study under multi-turn scenarios. The evaluation results consistently confirm that the proposed framework improves the consistency, clarity, and usefulness of the responses. These findings highlight the potential of memory-augmented LLMs for scalable and intelligent learning support within university environments. Full article

Driver distraction is a key factor contributing to traffic accidents. However, in existing computer vision-based methods for driver attention state recognition, monocular camera-based approaches often suffer from low accuracy, while multi-sensor data fusion techniques are compromised by poor real-time performance. To address these limitations, this paper proposes a Real-time Driver Attention State Recognition method (RT-DASR). RT-DASR comprises two core components: Binocular Vision Depth-Compensated Head Pose Estimation (BV-DHPE) and Multi-source Temporal Bidirectional Long Short-Term Memory (MSTBi-LSTM). BV-DHPE employs binocular cameras and YOLO11n (You Only Look Once) Pose to locate facial landmarks, calculating spatial distances via binocular disparity to compensate for monocular depth deficiency for accurate pose estimation. MSTBi-LSTM utilizes a lightweight Bidirectional Long Short-Term Memory (Bi-LSTM) network to fuse head pose angles, real-time vehicle speed, and gaze region semantics, bidirectionally extracting temporal features for continuous attention state discrimination. Evaluated under challenging conditions (e.g., illumination changes, occlusion), BV-DHPE achieved 44.7% reduction in head pose Mean Absolute Error (MAE) compared to monocular vision methods. RT-DASR achieved 90.4% attention recognition accuracy with 21.5 ms average latency when deployed on NVIDIA Jetson Orin. Real-world driving scenario tests confirm that the proposed method provides a high-precision, low-latency attention state recognition solution for enhancing the safety of mining vehicle drivers. RT-DASR can be integrated into advanced driver assistance systems to enable proactive accident prevention. Full article

Energy storage systems are becoming increasingly important as more renewable energy systems are integrated into the electrical (or power utility) grid. Low-cost and reliable energy storage is paramount if renewable energy systems are to be increasingly integrated into the power grid. Lead-acid batteries are widely used as energy storage for stationary renewable energy systems and agriculture due to their low cost, especially compared to lithium-ion batteries (LIB). However, lead-acid battery technology suffers from system degradation and a relatively short lifetime, largely due to its charging/discharging cycles. In the present study, we use Machine Learning methodology to estimate the battery degradation in an energy storage system. It uses two types of datasets: discharge condition and lead acid battery data. In the initial analysis, the Support Vector Regression (SVR) method with the RBF kernel showed poor results, with a low accuracy value of 0.0127 and RMSE 5377. On the other hand, the Long Short-Term Memory (LSTM) method demonstrated better estimation results with an RMSE value of 0.0688, which is relatively close to 0. Full article

Accurate data on solar photovoltaic (PV) generation is essential for the effective prediction of energy production and the effective management of distributed energy resources (DERs). Such data also plays a crucial role in ensuring the operation of DERs within modern power distribution systems is both safe and economical. Missing values, which may be attributed to faults in sensors, communication failures or environmental disturbances, represent a significant challenge for distribution system operators (DSOs) in terms of performing state estimation, optimal dispatch, and voltage regulation. This paper proposes a Pattern-Aware Bidirectional Long Short-Term Memory (PA-BiLSTM) model for solar generation imputation to address this challenge. In contrast to conventional convolution-based approaches such as the Convolutional Autoencoder and U-Net, the proposed framework integrates a 1D convolutional module to capture local temporal patterns with a bidirectional recurrent architecture to model long-term dependencies. The model was evaluated in realistic block–random missing scenarios (1 h, 2 h, 3 h, and 4 h gaps) using 5 min resolution PV data from 50 sites across 11 regions in South Korea. The numerical results show that the PA-BiLSTM model consistently outperforms the baseline methods. For example, with a time gap of one hour, it achieves an MAE of 0.0123, an R² value of 0.98, and an average MSE, with a maximum reduction of around 15%, compared to baseline models. Even under 4 h gaps, the model maintains robust accuracy (MAE = 0.070, R² = 0.66). The results of this study provide robust evidence that accurate, pattern-aware imputation is a significant enabling technology for DER-centric distribution system operations, thereby ensuring more reliable grid monitoring and control. Full article

As the deployment of wind energy systems continues to rise globally, ensuring the reliability and efficiency of grid-connected Doubly Fed Induction Generator (DFIG) wind turbines has become increasingly critical. Two core challenges faced by these systems include fault diagnosis in power electronic converters and accurate prediction of wind conditions for adaptive power control. Recent advancements in artificial intelligence (AI) have introduced powerful tools for addressing these challenges. This study presents the first unified comparative performance analysis of two deep learning-based models: (i) a Convolutional Neural Network-Long Short-Term Memory CNN-LSTM with Variational Mode Decomposition for real-time Grid Side Converter (GSC) fault diagnosis, and (ii) an Incremental Generative Adversarial Network (IGAN) for wind attribute prediction and adaptive droop gain control, applied to grid-integrated DFIG wind turbines. Unlike prior studies that address fault diagnosis and wind forecasting separately, both models are evaluated within a common MATLAB/Simulink framework using identical wind profiles, disturbances, and system parameters, ensuring fair and reproducible benchmarking. Beyond accuracy, the analysis incorporates multi-dimensional performance metrics such as inference latency, robustness to disturbances, scalability, and computational efficiency, offering a more holistic assessment than prior work. The results reveal complementary strengths: the CNN-LSTM achieves 88% accuracy with 15 ms detection latency for converter faults, while the IGAN delivers more than 95% prediction accuracy and enhances frequency stability by 18%. Comparative analysis shows that while the CNN-LSTM model is highly suitable for rapid fault localization and maintenance planning, the IGAN model excels in predictive control and grid performance optimization. Unlike prior studies, this work establishes the first direct comparative framework for diagnostic and predictive AI models in DFIG systems, providing novel insights into their complementary strengths and practical deployment trade-offs. This dual evaluation lays the groundwork for hybrid two-tier AI frameworks in smart wind energy systems. By establishing a reproducible methodology and highlighting practical deployment trade-offs, this study offers valuable guidance for researchers and practitioners seeking explainable, adaptive, and computationally efficient AI solutions for next-generation renewable energy integration. Full article

This work proposes a design technique to facilitate the design and optimization of a highperformance power amplifier (PA) in an automated manner. The proposed optimizationoriented strategy consists of the implementation of four deep neural networks (DNNs), sequentially. Firstly, a bidirectional long short-term memory (BiLSTM)-based DNN is trained based on the X-parameters for which the hyperparameters are optimized through the multi-objective ant lion optimizer (MOALO) algorithm. This step is significant since it conforms to the hidden-layer construction of DNNs that will be trained in the following steps. Afterward, a generative adversarial network (GAN) is employed for forecasting the load–pull contours on the Smith chart, such as gate and drain impedances that are employed for the topology construction of the PA. In the third phase, the classification the BiLSTM-based DNN is trained for the employed high-electron-mobility transistor (HEMT), leading to the selection of the optimal configuration of the PA. Finally, a regression BiLSTMbased DNN is executed, leading to optimizing the PA in terms of power gain, efficiency, and output power by predicting the optimal design parameters. The proposed method is fully automated and leads to generating a valid PA configuration for the determined transistor model with much more precision in comparison with long short-term memory (LSTM)-based networks. To validate the effectiveness of the proposed method, it is employed for designing and optimizing a PA operating from 1.8 GHz up to 2.2 GHz at 40 dBm output power. Full article

Long Short-Term Memory (LSTM) networks have revolutionized the field of deep learning, particularly in applications that require the modeling of sequential data. Originally designed to overcome the limitations of traditional recurrent neural networks (RNNs), LSTMs effectively capture long-range dependencies in sequences, making them suitable for a wide array of tasks. This survey aims to provide a comprehensive overview of LSTM architectures, detailing their unique components, such as cell states and gating mechanisms, which facilitate the retention and modulation of information over time. We delve into the various applications of LSTMs across multiple domains, including the following: natural language processing (NLP), where they are employed for language modeling, machine translation, and sentiment analysis; time series analysis, where they play a critical role in forecasting tasks; and speech recognition, significantly enhancing the accuracy of automated systems. By examining these applications, we illustrate the versatility and robustness of LSTMs in handling complex data types. Additionally, we explore several notable variants and improvements of the standard LSTM architecture, such as Bidirectional LSTMs, which enhance context understanding, and Stacked LSTMs, which increase model capacity. We also discuss the integration of Attention Mechanisms with LSTMs, which have further advanced their performance in various tasks. Despite their strengths, LSTMs face several challenges, including high Computational Complexity, extensive Data Requirements, and difficulties in training, which can hinder their practical implementation. This survey addresses these limitations and provides insights into ongoing research aimed at mitigating these issues. In conclusion, we highlight recent advances in LSTM research and propose potential future directions that could lead to enhanced performance and broader applicability of LSTM networks. This survey serves as a foundational resource for researchers and practitioners seeking to understand the current landscape of LSTM technology and its future trajectory. Full article

Unmanned aerial vehicles (UAVs) often face significant challenges in trajectory tracking within complex dynamic environments, where uncertainties, external disturbances, and nonlinear dynamics hinder accurate and stable control. To address this issue, a bio-inspired deep reinforcement learning (DRL) algorithm is proposed, integrating behavior cloning (BC) and long short-term memory (LSTM) networks. This method can achieve autonomous learning of high-precision control policy without establishing an accurate system dynamics model. Motivated by the memory and prediction functions of biological neural systems, an LSTM module is embedded into the policy network of the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm. This structure captures temporal state patterns more effectively, enhancing adaptability to trajectory variations and resilience to delays or disturbances. Compared to memoryless networks, the LSTM-based design better replicates biological time-series processing, improving tracking stability and accuracy. In addition, behavior cloning is employed to pre-train the DRL policy using expert demonstrations, mimicking the way animals learn from observation. This biomimetic plausible initialization accelerates convergence by reducing inefficient early-stage exploration. By combining offline imitation with online learning, the TD3-LSTM-BC framework balances expert guidance and adaptive optimization, analogous to innate and experience-based learning in nature. Simulation experimental results confirm the superior robustness and tracking accuracy of the proposed method, demonstrating its potential as a control solution for autonomous UAVs. Full article

This study proposes a Compressive Sensing Convolutional Long Short-Term Memory (CSCL) model that aims to improve short-term (12–24 h) forecast accuracy compared to standard ConvLSTM. It is especially useful when subtle spatiotemporal variations complicate feature extraction. CSCL uses uniform sampling to partially mask spatiotemporal wave fields. The model training strategy integrates both complete and masked samples from pre- and post-sampling. This design encourages the network to learn and amplify subtle distributional differences. Consequently, small variations in convolutional responses become more informative for feature extraction. We considered the theoretical explanations for why this sampling-augmented training enhances sensitivity to minor signals and validated the approach experimentally. For the region 120–140° E and 20–40° N, a four-layer CSCL model using the first five moments as inputs achieved the best prediction performance. Compared to ConvLSTM, the R² for significant wave height improved by 2.2–43.8% and for mean wave period by 3.7–22.3%. A wave-energy case study confirmed the model’s practicality. CSCL may be extended to the prediction of extreme events (e.g., typhoons, tsunamis) and other oceanic variables such as wind, sea-surface pressure, and temperature. Full article

As investors increasingly use Environmental, Social, and Governance (ESG) criteria, a key challenge remains: ESG data is typically reported annually, while financial markets move much faster. This study investigates whether incorporating annual ESG scores can improve monthly stock return forecasts for German DAX-listed firms. We employ a multivariate long short-term memory (LSTM) network, a machine learning model ideal for time series data, to test this hypothesis over two periods: an 8-year analysis with a full set of ESG scores and a 16-year analysis with a single disclosure score. The evaluation of model performance utilizes standard error metrics and directional accuracy, while statistical significance is assessed through paired statistical tests and the Diebold–Mariano test. Furthermore, we employ SHapley Additive exPlanations (SHAP) to ensure model explainability. We observe no statistically significant indication that incorporating annual ESG data enhances forecast accuracy. The 8-year study indicates that using a comprehensive ESG feature set results in a statistically significant increase in forecast error (RMSE and MAE) compared to a baseline model that utilizes solely historical returns. The ESG-enhanced model demonstrates no significant performance disparity compared to the baseline across the 16-year investigation. Our findings indicate that within the one-month-ahead projection horizon, the informative value of low-frequency ESG data is either fully incorporated into the market or is concealed by the significant forecasting capability of the historical return series. This study’s primary contribution is to demonstrate, through out-of-sample testing, that standard annual ESG information holds little practical value for generating predictive alpha, urging investors to seek more timely, alternative data sources. Full article

The rapid development of wireless communication technology is leading to increasingly scarce spectrum resources, making efficient utilization a critical challenge. This paper proposes a Convolutional Neural Network–Long Short-Term Memory-Integrated Gradient-Weighted Class Activation Mapping (GC-CNN-LSTM) model, aimed at enhancing the accuracy of long-term spectrum prediction across multiple frequency bands and improving model interpretability. First, we achieve multi-frequency long-term spectrum prediction using a CNN-LSTM and compare its performance against models including LSTM, GRU, CNN, Transformer, and CNN-LSTM-Attention. Next, we use an improved Grad-CAM method to explain the model and obtain global heatmaps in the time–frequency domain. Finally, based on these interpretable results, we optimize the input data by selecting high-importance frequency points and removing low-importance time segments, thereby enhancing prediction accuracy. The simulation results show that the Grad-CAM-based approach achieves good interpretability, reducing RMSE and MAPE by 6.22% and 4.25%, respectively, compared to CNN-LSTM, while a similar optimization using SHapley Additive exPlanations (SHAP) achieves reductions of 0.86% and 3.55%. Full article

The main objective of the paper is to verify whether the integration of attention mechanisms could improve the effectiveness of online fake news detection models. The models were training using selected deep learning methods, which were suitable for text processing, such as CNN (Convolutional Neural Network), LSTM (Lon-short Term Memory), BiLSTM (Bidirectional LSTM), GRU (Gated Recurrent Unit), and transformer. The novelty of the paper lies in the addition of attention mechanisms to each of those models, and comparison of their performance across both datasets, LIAR and WELFake. Afterwards, an analysis of resulting changes in terms of the detection performance was carried out. The paper also describes the issue of toxicity in the online space and how it affects society, the toxicity sources, and methods to tackle it. Furthermore, the article provides a description of individual deep learning methods and the principles of attention mechanism. Finally, it was shown that the attention mechanism can increase the accuracy of basic models for fake news detection; however, the differences are insignificant in the case of the LIAR dataset. The reason for this can be found in the dataset itself. In contrast, the addition of attention mechanism to models on the WELFake dataset showed a significant improvement of results, where the average accuracy was 0.967 and average F1-rate was 0.968. These results were better than the results of experiments with the simple transformer. Comparison of the results showed that it makes sense to enrich the basic neural network models with the attention mechanisms, especially with the multi-head attention mechanism. The key finding is that attention mechanisms can enhance fake news detection performance when applied to high-quality, well-balanced datasets. Full article

Accurate estimation of the State-of-Health (SOH) of lithium-ion batteries is crucial for ensuring the safety and longevity of electric vehicles and energy storage systems. However, conventional LSTM models often fail to capture the nonlinear degradation dynamics and long-term dependencies of battery aging. This study proposes a Fractional-Derivative Enhanced LSTM (F-LSTM), which incorporates fractional parameters α and Δt into the cell state update to model multi-scale memory effects. Experiments conducted on the CALCE LiCoO₂ dataset and the Tongji University NCA dataset demonstrate that, compared with the standard LSTM, the proposed F-LSTM reduces RMSE and MAE by more than 40% while maintaining robust performance across different chemistries, temperatures, and dynamic conditions. These results highlight the potential of integrating fractional calculus with deep learning to achieve accurate SOH prediction and support intelligent battery management. Full article

During the continuous operation of trailing suction hopper dredger (TSHD), equipment workload exhibits significant time-varying characteristics. Maintaining dynamic symmetry between power generation and consumption is crucial for ensuring system stability and preventing power supply failures. Key challenges lie in dynamic perception, accurate prediction, and real-time power management to achieve this equilibrium. To address this issue, this paper proposes and constructs a “prediction-driven dynamic power management method.” Firstly, to model the complex temporal dependencies of the workload sequence, we introduce and improve a dilated convolutional long short-term memory network (Dilated-LSTM) to build a workload prediction model with strong long-term dependency awareness. This model significantly improves the accuracy of workload trend prediction. Based on the accurate prediction results, a dynamic power management strategy is developed: when the predicted total power consumption is about to exceed a preset margin threshold, the Power Management System (PMS) automatically triggers power reduction operations for adjusfigure loads, aiming to maintain grid balance without interrupting critical loads. If the power that the generator can produce is still less than the required power after the power is reduced, and there is still a risk of supply-demand imbalance, the system uses an Improved Grey Wolf Optimization (IGWO) algorithm to automatically disconnect some non-critical loads, achieving real-time dynamic symmetry matching of generation capacity and load demand. Experimental results show that this mechanism effectively prevents generator overloads or ship-wide power failures, significantly improving system stability and the reliability of power supply to critical loads. The research results provide effective technical support for intelligent energy efficiency management and safe operation of TSHDs and other vessels with complex working conditions. Full article

This study develops a ship propeller diagnostic system to address the issue of insufficient ship maintenance capacity and enhance operational efficiency. It uses the Remaining Useful Life (RUL) prediction technology to establish a sensing platform for ship propellers to capture vibration signals during ship operations. The Diagnosis and RUL Prediction Model is designed to assess bearing aging status and the RUL of the propeller. The synchronized signal reshaping technology is employed in the Diagnosis and RUL Prediction Model to process the original vibration signals as input to the model. The vibration signals obtained are used to analyze the temporal and spectral energy of propeller bearings. Exponential functions are used to generate the health index as model outputs. Model inputs and outputs are simultaneously input into a Long Short-Term Memory (LSTM) model for training, culminating as the Diagnosis and RUL Prediction Model. Compared to Recurrent Neural Network and Support Vector Regression models used in previous studies, the Diagnosis and RUL Prediction Model developed in this study achieves a Mean Squared Error (MSE) of 0.018 and a Mean Absolute Error (MAE) of 0.039, demonstrating outstanding performance in prediction results and computational efficiency. This study integrates the Diagnosis and RUL Prediction Model, bearing aging experimental data, and real-world vibration measurements to develop the diagnosis and RUL prediction system for ship propellers. Experiments with ship propellers show that when the bearing of the propeller enters the worn stage, this diagnostic system for ship propellers can accurately determine the current status of the bearing and its remaining useful life. This study offers a practical solution to insufficient ship maintenance capacity and contributes to improving the operational efficiency of ships. Full article