Artificial Intelligence-Based Electric Vehicle Smart Charging System in Malaysia

Abstract

The worldwide transition to electric vehicles (EVs) is gaining momentum, propelled by the imperative to reduce carbon emissions and foster sustainable transportation. In Malaysia, the government is facilitating this transformation through targeted initiatives aimed at promoting the use of electric vehicles (EVs) and developing the required infrastructure. This paper investigates the crucial role of artificial intelligence (AI) in developing intelligent electric vehicle (EV) charging infrastructure, specifically focusing on the context of Malaysia. The paper examines the current electric vehicle (EV) charging infrastructure in Malaysia, highlights advancements led by artificial intelligence (AI), and references both local and international case studies. Fluctuations in the Total Industry Volume (TIV) and Total Industry Production (TIP) reflect changes in market demand and production capabilities, with notable peaks in March 2023 and March 2024. The research reveals that AI technologies, such as machine learning and predictive analytics, can enhance charging efficiency, improve user experience, and support grid stability. A mathematical model for an AI-based smart charging system was developed, and the implemented system achieved 30% energy savings and a 20.38% reduction in costs compared to traditional methods. These findings underscore the system’s energy and cost efficiency. In addition, we outline the potential advantages and challenges associated with incorporating artificial intelligence (AI) into Malaysia’s electric vehicle (EV) charging infrastructure. Furthermore, we offer recommendations for researchers, industry stakeholders, and regulators. Malaysia can enhance the uptake of electric vehicles and make a positive impact on the environment by leveraging artificial intelligence (AI) to enhance its electric vehicle charging system (EVCS).

1. Introduction

Nations are rapidly transforming their transport systems to combat climate change and reduce carbon emissions. Electric vehicles (EVs) are central to this shift, offering a sustainable alternative to conventional cars [1]. As EV usage grows, so does the need for efficient charging infrastructure. Traditional charging systems, with fixed schedules and constant rates, often fail to adapt to the dynamic needs of renewable energy and EV users [2]. Advanced charging systems, leveraging cutting-edge technology, aim to address these challenges by improving energy efficiency and customer satisfaction. The surge in EV adoption necessitates advanced charging infrastructure that caters to diverse consumer needs and supports the power grid. Smart charging systems optimize energy use by adjusting schedules based on real-time data [3]. These systems reduce costs by charging during low-demand periods and avoiding peak demand charges, offering significant savings. They also enhance user experience by allowing personalized charging, real-time status monitoring, and notifications. Crucially, smart charging systems help maintain grid stability [4], balancing loads, preventing overloads, and integrating renewable energy sources.

These systems also support Vehicle-to-Grid (V2G) technology, enabling bidirectional energy flow between EVs and the grid [5]. V2G systems provide incentives for EV owners and assist the grid during peak demand. Moreover, smart charging systems generate valuable data on charging patterns and energy use, aiding future technology and policy improvements. The specifications of EV charging systems are shown in Figure 1.

In Malaysia, the EV market is emerging, with Battery Electric Vehicles (BEVs) reducing greenhouse gas emissions and fossil fuel use [6]. The government supports BEVs through initiatives like the New Energy Vehicle (NEV) project and the Energy Efficient Vehicle (EEV) Roadmap [7]. The Low Carbon Mobility Blueprint 2021–2030 aims for 100,000 EVs and 4000 EV buses by 2030, as depicted in Figure 2.

Malaysia’s BEV sector faces challenges, including infrastructure and costs. A diverse approach, involving government support, public-private partnerships, and consumer education, is essential for growth. The Low-Carbon Mobility Blueprint (LCMB) 2021–2030 plans to build 10,000 EV charging stations by 2025, including 1000 DC fast chargers and 9000 AC chargers [8]. Understanding the BEV market’s characteristics and challenges is crucial. Integrating EV charging stations into the grid requires advanced AI technologies for network inspection and control, improving efficiency and stability. Globally, the EV market is growing rapidly. In 2020, 3 million new EVs were delivered, and by 2030, 145 million EVs are projected, with BEVs comprising 67% of EVs [9]. The global sales of EVs are illustrated in Figure 3.

In Malaysia, EV sales lag behind sales of traditional vehicles, but interest in EVs is growing due to government subsidies, environmental awareness, and technological advancements. The global light car market showed a +10.5% growth, and although EV adoption is influenced by broader automotive trends, environmental activism and financial incentives are significant drivers. In Malaysia, EV sales surged by 286% to 10,159 units in 2023, while hybrid vehicle sales increased by 40% [9]. EV sales in Malaysia are shown in Figure 4.

Total industry volume (TIV) 1H2024 versus 1H2023 by month, and the total industry production (TIP) trend 1H2024 versus 1H2023 by month, are shown in Figure 5 and Figure 6, respectively. Both TIV and TIP for the first half of 2024 show a generally positive trend compared to the same period in 2023, indicating growth in the EV market in Malaysia. There are noticeable fluctuations in both TIV and TIP on a month-to-month basis, reflecting possible variations in market demand, production capacity, and external factors affecting the EV industry. March 2023 recorded the highest TIV, while March 2024 also saw a peak but at a lower volume, highlighting a significant sales surge during this month in 2023 [10].

Malaysia’s status as a leading nickel producer significantly impacts the EV market by providing a crucial raw material for lithium-ion batteries, essential for electric vehicle (EV) propulsion. Nickel’s role in increasing battery energy density and performance positions Malaysia strategically within the global EV supply chain. This local supply advantage can help stabilize material costs and reduce battery production expenses, potentially lowering EV prices and accelerating adoption. Furthermore, Malaysia’s nickel resources may attract investment in battery manufacturing and processing facilities, bolstering the country’s EV infrastructure and technological capabilities. As global EV demand grows, Malaysia’s ability to supply high-quality nickel supports both economic development and sustainable practices, positioning the nation favorably in the transition to electric mobility.

Although they are still in their infancy, artificial intelligence (AI) applications in Malaysian EV charging stations have great potential to improve the nation’s infrastructure for charging vehicles. While there has not been much widespread implementation, AI technologies like machine learning and predictive analytics are being utilized to improve several parts of the ecosystem around EV charging. AI is already being used in demand forecasting applications to assist anticipate peak billing periods by utilizing user behavior and previous data. This lowers the possibility of blackouts by enabling improved control over the distribution of electricity and guaranteeing that the grid can support the load during times of high demand. Another use is in dynamic pricing and energy management, where AI algorithms modify charging rates in real time according to market variables, grid circumstances, and energy availability to guarantee grid stability and cost-effectiveness. Furthermore, by guiding cars to less crowded stations or more effectively arranging charging sessions, AI is being incorporated into charging station management systems to optimize the distribution of charging slots and lessen traffic.

Malaysia still has difficulties expanding AI-driven solutions throughout its EV charging network, despite these advancements. Important obstacles include the lack of a sufficient charging infrastructure, expensive startup costs, and the requirement for more localized data for AI model fine-tuning. Ongoing pilot programs and partnerships between public and commercial organizations, however, are laying the foundation for future, more thorough AI integration.

Understanding the trends and challenges in EV adoption and the role of AI in optimizing smart charging systems is crucial for ensuring a sustainable and efficient transition to electric mobility in Malaysia and globally. The rest of the paper is organized as follows: Section 2 discuss the overview of EV charging systems along with the progression of EV charging systems. AI-driven innovations in the Malaysian EV charging system and future prospects and recommendations are portrayed in Section 4 and Section 5, respectively. In conclusion, final observations and suggestions for future work are given in Section 6.

2. Overview of EV Charging Systems

Electric vehicles (EVs) need efficient and dependable charging infrastructure to become widely used. Charging stations are usually classed by power. Level 1 charging uses 120 V outlets and is sluggish but adequate for overnight home usage. Level 2 chargers, which utilize 240 V, charge more quickly and are prevalent in homes, offices, and public venues. Long-distance travel requires DC fast chargers, or Level 3, which are frequently found along highways and in business locations. AI and communication technologies optimize charging schedules depending on power costs and grid demand, boosting cost-effectiveness and grid stability beyond hardware. Three primary sorts of charging facilities are identified based on the location of the charging infrastructure. Each of these facilities has distinct attributes and varying degrees of charge. Figure 7 provides a comprehensive overview of the categories of infrastructure.

2.1. Conductive-Based EV Charging System

Conductive charging was the foundation of the first electric vehicle charging infrastructure. Electric vehicles are able to be charged using this charging infrastructure. Garages, businesses, and some outdoor spaces are common places to find them. In order to get a good charge out of this model’s battery, it takes a lengthy time to charge it. Also, the bigger battery capacities needed for EVs charged this way add extra weight and expense to the EV. The two main categories of electric vehicles that employ conductive charging are battery-electric and plug-in hybrids.

There are two types of conductive charging: AC and DC. There are two tiers to the AC charging infrastructure, and a third is in the works. Currently, DC level 3 chargers are the norm for rapid charging. It is believed that the Model 3 can be charged at a pace of up to 1000 miles per hour using Tesla’s DC level 3 charger, the supercharger [11]. Multiple DC fast charging protocols exist, including GB/T, CHAdeMO, and the combined charging system (CCS), in addition to the Tesla supercharger [12,13]. Japan uses the CHAdeMO standard for DC rapid charging, while China uses the GB/T standard. Around the globe, including in the United States and Europe, the CCS serves as the de facto standard. Still, a growing number of areas outside of their primary regions are adopting these distinct norms. Figure 8 shows the different charging system connections.

2.2. Wireless-Based EV Charging System

Numerous studies aim to improve the electric vehicle (EV) experience. A key focus is wireless charging to address range anxiety and reduce EV operating costs. Research on wireless EV charging can be divided into optimization-based and non-optimization-based models. Optimization-based models often target route scheduling and infrastructure placement. For instance, Zhang et al. [14] proposed a charging scheme to minimize operational costs for electric buses with wireless charging, using speed assumptions to develop a scheduling algorithm. Other models, like [15,16], use particle swarm optimization to balance battery and infrastructure costs in a multi-road network. However, these models can be complex, with assumptions such as fully charged batteries at the start of a journey. Another model in [17] reduces the number of charging stations while considering power flow constraints, though simplified linear models may affect accuracy.

Static wireless charging involves charging EVs wirelessly while parked near charging stations [18]. This method is convenient for locations like parking lots but requires substantial battery capacity, increasing EV costs. INDUCTEV has developed a 300 kW dynamic wireless charging system with 90% efficiency [19], but it comes with high installation costs [20] and faces challenges with power transfer efficiency and maintenance [21]. The Drop & Charge system from Humavox [22] simplifies charging by allowing devices to charge when placed on a key-shaped station. It uses two coils for efficient energy transmission through magnetic coupling, a common method in wireless charging for devices like phones and laptops [22], which is shown in Figure 9.

2.3. Battery Swapping System

At battery swapping systems (BSSs), users are able to charge their empty batteries and then change them out for full ones. Over time, a fully charged battery will supplant the EV battery. Buses powered by huge batteries that take a long time to charge using traditional conductive charging methods could use battery swapping. A large supply of this technology must be maintained by the BSS or a third party that provides electric vehicle owners with borrowed batteries [23]. Powering the BSS are a distribution transformer, batteries, battery switching gear, and AC/DC converters for battery charging. Some studies have shown that BSS may use V2G paradigm power services that include bidirectional pricing. Significant infrastructure expenditures, a large BSS footprint, and battery consistency are challenges for this system. In 2013, Tesla Company introduced a battery swapping system that would allow for a battery change in 90 s [24]. Comparison of different EV charging systems is shown in

Table 1. Comparison of EV Charging Systems.

Charging System	Advantages	Disadvantages	Solutions
Conductive-Based EV Charging System	- High efficiency and fast charging speeds. - Widely available with established infrastructure. - Lower cost compared to wireless systems.	- Requires physical connection, which can be inconvenient for users. - Susceptible to wear and tear over time.	- Develop automated connection systems to improve user convenience. - Enhance durability of connectors to reduce maintenance needs.
Wireless-Based EV Charging System	- Convenient and user-friendly, with no need for physical connection. - Reduced wear and tear due to lack of physical connectors.	- Lower efficiency compared to conductive charging. - Higher cost and more complex infrastructure. - Potential for electromagnetic interference.	- Improve efficiency through advanced coil design and resonant frequency optimization. - Implement shielding techniques to minimize interference.
Battery Swapping System	- Quick energy replenishment by swapping depleted batteries with fully charged ones. - Reduces wait time compared to traditional charging.	- Requires standardized batteries, which limits vehicle design flexibility. - High initial setup cost for swapping stations. - Logistic challenges in battery management.	- Promote standardization across manufacturers. - Develop efficient logistics and battery management systems to reduce operational costs.

.

2.4. Development of EV Charging Systems with AI

The development of electric vehicle (EV) charging systems has evolved significantly with technology and AI integration. Initially, from the late 19th century to the mid-20th century, early EVs used basic home electrical outlets for charging. The oil crisis of the 1970s revived interest in EVs and led to improvements in charging systems, though public infrastructure remained sparse. In the 1990s, early models like the GM EV1 faced range anxiety and limited charging options. The introduction of standardized connectors such as SAE J1772 in the 2000s improved compatibility, and protocols like CHAdeMO (2009) and the Combined Charging System (CCS) facilitated faster DC charging [25].

In the 2010s, the EV market expanded rapidly with models like the Nissan Leaf and Tesla’s offerings gaining popularity. Tesla notably introduced its Supercharger network, pioneering high-speed DC fast-charging capabilities. This decade also witnessed the integration of AI into EV charging systems. AI algorithms began optimizing charging schedules based on factors such as energy demand, grid capacity, and user preferences. Machine learning techniques enabled predictive maintenance of charging infrastructure, ensuring reliability and efficiency. Entering the 2020s, advancements in ultra-fast charging technologies, capable of delivering up to 350 kW, promised to significantly reduce charging times, approaching the convenience of refueling traditional vehicles. AI continued to play a crucial role in shaping the future of EV charging. Smart charging systems leveraged AI and the IoT (Internet of Things) to manage charging sessions dynamically, optimizing energy usage and minimizing grid impact [26].

Moreover, the concept of Vehicle-to-Grid (V2G) technology emerged, allowing EVs to not only consume electricity but also to return power to the grid during peak demand periods, enhancing grid stability and resilience. A complete list of V2G pilot projects around the world can be found in [27]. As shown in

Table 2. Summary of some significant V2G pilot projects worldwide.

Project Name	No. of Chargers	Service	Timespan	Country
SunnYparc	250	Time shifting, pricing scheme testing, reserve	2022–2025	Switzerland
Electric Heavy Goods Vehicles—first roll-out and demonstration of V2X and grid decarbonization	unknown	Freq. Response, Dist. Services	2024–ongoing	Great Britain
SmartMAUI, Hawaii	80	Distribution services, frequency response, time shifting	2012–2015	US
V2X Suisse	40	Freq. Response, Reserve, Arbitrage, Dist. Services	2021–2023	China
Leaf to home	4000	Emergency backup, time shifting	2012–ongoing	Japan
Utrecht V2G charge hubs	80	Arbitrage	2018–ongoing	Netherlands
Bidirektionales Lademanagement—BDL	50	Arbitrage, frequency response, time shifting	2021–2022	Germany
Fiat-Chrysler V2G	600	Load balancing	2019–2021	Italy
UK Vehicle-2-Grid (V2G)	100	Support power grid	2016–ongoing	UK
INVENT—UCSD/Nissan/Nuvve	50	Distribution services, frequency response, time shifting	2017–2020	US
Share the Sun/Deeldezon Project	80	Distribution services, frequency response, time shifting	2019–2021	Netherlands
VGI core comp. dev. and V2G demo. using CC1	100	Arbitrage, frequency response, reserve, time shifting	2018–2022	South Korea
Electric Nation Vehicle to Grid	100	Distribution services, reserve, time shifting	2020–2022	UK
OVO Energy V2G	320	Arbitrage	2018–2021	UK
Powerloop: Domestic V2G Demonstrator Project	135	Arbitrage, distribution services, emergency backup, time shifting	2018–ongoing	UK
Realizing Electric Vehicle to Grid Services	51	Frequency response, reserve	2020–2022	Australia
Parker	50	Arbitrage, distribution services, frequency response	2016–2018	Denmark
I-GReta	1	Reserve, Dist. Services, Time shifting, Emergency back up	2021–2023	Astria
INCIT-EV	100	Freq. Response, Reserve, Dist. Services, Time shifting, Emergency back up	2020–2024	Spain

, most of the V2G pilot projects were only initiated in recent years, and the scale of these pilot projects is small. Moreover, most of the existing literature only focuses on the EV charging aspect. AI-driven analytics enabled real-time data analysis, supporting grid operators and EV owners in making informed decisions about charging behavior and energy consumption patterns. Looking ahead, the integration of AI is set to further evolve EV charging systems, driving innovations in wireless charging technologies, enhancing interoperability across charging networks, and facilitating seamless integration with renewable energy sources. As governments and industries continue to invest in infrastructure and technology, AI will undoubtedly play a pivotal role in shaping a sustainable and efficient future for electric mobility.

AI technology has transformed electric vehicle (EV) charging systems, improving user experience, reliability, and efficiency. AI algorithms improve charging plans using real-time power use, grid capacity, and user behavior. This allows EVs to charge at the ideal periods, when renewable energy is abundant and electricity prices are low, reducing costs and environmental effects. Machine learning-based predictive maintenance monitors charging infrastructure for problems and inefficiencies before they occur, reducing downtime and improving reliability [28]. AI-driven smart grids may integrate EVs into energy management plans to balance system loads and participate in demand response. AI provides real-time charging station availability, specialized charging suggestions, and dynamic pricing schemes that incentivize off-peak charging, improving customer convenience. These discoveries demonstrate AI’s importance in EV adoption and sustainable mobility solutions worldwide. The limitations of AI in EV charging systems along with potential solutions is shown in

Table 3. Limitations of AI in EV charging systems along with potential solutions.

Limitation	Description	Solution
Data Quality and Availability	AI models require high-quality, large datasets for accurate predictions. Inconsistent or incomplete data can lead to suboptimal performance.	Implement data augmentation techniques, use synthetic data generation, and establish standardized data collection protocols.
Scalability Issues	Many AI algorithms struggle to scale effectively across large numbers of EVs or charging stations, leading to performance bottlenecks.	Optimize algorithms for parallel processing and distributed computing, and integrate cloud-based solutions.
Real-Time Processing Constraints	AI models need to process data and make decisions in real time, but latency and computational overhead can impede timely responses.	Employ edge computing to reduce latency, and optimize AI models for faster computation and reduced resource requirements.
Integration with Existing Infrastructure	Integrating AI with legacy systems and infrastructure can be complex and costly, leading to delays in deployment.	Develop modular AI systems with flexible APIs for easier integration, and offer phased implementation strategies.
Cybersecurity Risks	AI-driven systems are vulnerable to cyber-attacks, which can compromise data integrity and system reliability.	Implement robust encryption, multi-factor authentication, and continuous monitoring for potential security breaches.
Interpretability and Transparency	AI models, especially deep learning, often act as “black boxes”, making it difficult to understand how decisions are made.	Develop explainable AI (XAI) models that provide insights into decision-making processes and improve user trust.
Cost of Implementation	High costs associated with AI technology deployment, including hardware, software, and skilled personnel, can be prohibitive.	Leverage open-source AI frameworks, focus on incremental deployment, and explore government subsidies and partnerships.
User Acceptance and Trust	Users may be reluctant to trust AI systems for critical functions like charging, especially if the benefits and reliability are unclear.	Increase user education and engagement, provide clear demonstrations of reliability, and offer user control options.
Adaptability to Changing Conditions	AI models may struggle to adapt to rapidly changing conditions such as fluctuating energy prices or sudden changes in grid demand.	Integrate adaptive learning algorithms that continuously update models based on new data and changing conditions.
Ethical and Regulatory Compliance	Ensuring AI systems comply with evolving regulations and ethical standards can be challenging, especially in diverse markets.	Develop AI systems with built-in compliance monitoring, and engage with policymakers to shape future regulations.

.

2.5. AI Optimization in Smart EV Charging Systems

The growing use of Battery Electric Vehicles (BEVs) in residential areas will impact power demand, necessitating detailed research on distribution network planning and load profiles. Uncontrolled high-power EV charging, like 11.1 kW and 22.2 kW, can overload the grid, as observed in Finnish studies [29]. To manage this demand, AI-based Demand Response (DR) algorithms are essential. Unlike traditional methods that involve costly and environmentally harmful infrastructure expansion, AI-based DR uses machine learning and advanced analytics to align EV charging with available supply [30]. In high-rise residential complexes with limited electrical capacity, AI DR systems help balance EV charging and prevent grid strain. By adjusting charging schedules based on driver behavior, these systems can lower greenhouse gas emissions, reduce power costs, and enhance system stability. Optimization algorithms such as Artificial Neural Networks (ANN), Dynamic Programming (DP), Fuzzy Logic (FL), Game Theory, and Particle Swarm Optimization (PSO) are used to minimize consumer costs, transformer loads, energy prices, network losses, voltage variations, and peak loads, improving the economic feasibility of EV integration [31].

Despite this, there is a gap in existing research concerning the prediction of load demand when incorporating EV driver behavior and electrical system capacity. Most studies do not simultaneously address these factors. The proposed research aims to fill this gap by developing control algorithms that consider these elements, suggesting capacity upgrades and the addition of EV chargers when necessary [32]. Although simulation results show that optimized charging schedules can mitigate power fluctuations, the models are limited by static assumptions about EV types and charging rates. Future research should integrate discharging features and variable charging rates to enhance power fluctuation control [33,34,35,36,37]. Power Fluctuation Level for different EV charging scheme is displayed in Figure 10.

Moreover, real-time data are essential for verifying the effectiveness of AI DR optimization algorithms. Without large datasets, it is difficult to validate these algorithms’ performance and safety. Real data from sources like smart meters and charging stations are necessary for accurate predictions and reliable system operation [38]. Additionally, integrating individual user behavior and dynamic factors with system capacities remains a challenge. Advanced AI algorithms like Deep Reinforcement Learning (DRL) and ANN show promise in addressing these issues, especially when considering discharging features and unpredictable charging rates [39,40,41].

Survey analysis from the Netherlands also provides insights into BEV driver behavior, revealing that route choice and charging behavior are significantly influenced by traditional route characteristics, vehicle-related factors, and charging features. These findings can guide the development of more effective DR strategies [42]. Non-Intrusive Load Extracting (NILE) algorithms and Latin Hypercube Sampling (LHS) models offer alternative approaches to managing uncertainties in EV charging behavior, though they have limitations, such as data sampling rates and a lack of DR control over limited electrical system capacities [43,44]. The extracted charging does not estimate the state of charge (SOC) of the car remaining. Lastly, the extracted information does not allow DR control over a limited electrical system capacity network. [44] suggests using the Latin Hypercube Sampling (LHS) model to deal with EVs arrival, departure time, and SOC origin uncertainties. That research has a similarity with the current research in that an algorithm is needed to determine the daily arrival, departure, and SOC uncertainties; however, the difference is that [44] uses LHS to determine the lowest charging cost, while this research emphasizes DR optimum charging algorithms.

Table 4. Research finding of AI optimization for EV smart charging systems.

Aspect	Research Summary	Research Gaps	Future Work	Ref.
Battery Management Systems (BMS)	Analysis of AI methods in BMS technology for EVs.	Lack of advanced predictive analytics and cybersecurity measures in BMS.	Enhance BMS algorithms, explore advanced predictive analytics, and improve cybersecurity.	[45]
Hybrid Energy Systems	Evaluation of hybrid energy systems and their impact on COE in India.	High infrastructure and maintenance costs, financial constraints, and technical challenges in hybrid systems.	Research technological challenges and develop economic pathways for hybrid systems.	[46]
EV Charging Frameworks	Proposed analytical framework for EV charging studies.	Absence of benchmarks or frameworks for EV charging control studies.	Establish benchmarks and frameworks to standardize methods and improve comparison of outcomes.	[47]
Health Prognosis for EVs	Analysis of Bayesian Inference for health prognosis of electric vehicles.	Limited integration of real-time data for enhanced predictive accuracy.	Integrate real-time data and refine the Bayesian Inference model to improve prediction accuracy.	[48]
State-of-Charge Estimation	Application of deep neural networks for SOC estimation of Li-ion batteries.	Requires more robust models for extreme conditions and faster computation.	Develop more robust models and optimize algorithms for faster computation under extreme conditions.	[49]
Battery State Estimation	Use of genetic algorithm-based fuzzy C-means for battery state estimation in EVs.	Lack of adaptability in rapidly changing environments.	Enhance adaptability of algorithms in dynamic environments with varying conditions.	[50]
State-of-Health Estimation	Support vector machines for SOH estimation of Li-ion batteries in EVs.	Limited by model generalization across different battery types.	Expand the generalization of models to accommodate a wider range of battery chemistries and types.	[51]
HESS Topologies	Wavelet transform approach for HESS topologies in EV batteries.	Insufficient analysis of long-term performance in diverse scenarios.	Conduct comprehensive long-term performance analysis across various operational scenarios.	[52]
State-of-Charge Estimation	Fuzzy neural networks and Elman neural networks for SOC estimation in EVs.	Complexity in algorithm implementation and scalability issues.	Simplify algorithm implementation and improve scalability for large-scale deployments.	[53]
SOC Estimation of Li-ion Batteries	Application of lightning search algorithm for SOC estimation in Li-ion batteries.	Lack of validation in diverse environmental conditions.	Validate the lightning search algorithm across a broader range of environmental conditions.	[54]
Safety Envelope for LIBs	Artificial neural networks for defining the safety envelope of lithium-ion batteries.	Gaps in real-time safety monitoring and response.	Implement real-time safety monitoring systems and enhance response mechanisms using AI.	[55]
Big Data-Driven Modeling	Extreme learning machine for big data-driven lithium-ion battery modeling.	Challenges in handling high-dimensional data efficiently.	Develop more efficient data processing techniques for high-dimensional data in battery modeling.	[56]
State of Charge Prediction	Gaussian process regression and linear regression models for SOC prediction of EV batteries.	Needs improved prediction accuracy under variable operating conditions.	Enhance prediction accuracy by incorporating more complex machine learning models and real-time data.	[57]
Battery Modeling and Management	Rain-flow cycle counting for aging-considered battery model and management.	Difficulty in accurately predicting battery degradation over long periods.	Develop more accurate long-term degradation prediction models using advanced machine learning techniques.	[58]
Battery Aging Dataset Generation	Supervised training algorithms for large-scale battery aging dataset generation.	Lack of diverse and comprehensive battery aging datasets.	Create more comprehensive and diverse battery aging datasets to improve model training and validation.	[59]
Battery Lifetime Prediction	Feature-based data-driven approach for battery lifetime prediction.	Challenge in accurately predicting battery lifetime across various use cases.	Expand the feature set and improve model generalization for different battery use cases.	[60]
EV Charging Management Systems	Reinforcement learning for EV charging management systems.	Limited scalability and adaptability in dynamic grid environments.	Enhance scalability and adaptability of reinforcement learning models for diverse grid conditions.	[61]
Energy Management of Multi-source EVs	Artificial neural network and convex optimization for energy management in multi-source EVs.	Computational complexity and high dependency on precise model parameters.	Simplify computational processes and reduce dependency on highly specific model parameters.	[62]
Hybrid Electric Vehicle Charging	Whale optimization algorithm for hybrid AC-DC microgrid in EV charging.	Inefficiencies in handling large-scale charging demands.	Optimize the whale optimization algorithm for better scalability and efficiency in large networks.	[63]
AI Algorithms for Optimization	AI algorithms (e.g., ML, DL) improve EV charging efficiency and reduce grid impact.	Limited integration with existing infrastructure.	Develop scalable AI models and conduct real-world trials.	[64]
Predictive Analytics	Enhances charging strategies by forecasting demand and user behavior.	Insufficient data on long-term user behavior.	Gather extensive user behavior data and validate with long-term studies.	[65]
User Behavior Consideration	Enhances personalization and user satisfaction by incorporating behavior.	Limited understanding of diverse user behaviors.	Conduct extensive user behavior studies.	[66]
Online State-of-Charge Estimation	Deep neural networks for online SOC estimation using electrochemical impedance spectroscopy.	Gaps in real-time model adaptability for online applications.	Improve real-time adaptability and accuracy for online state-of-charge estimations using dynamic models.	[67]
Hybrid Electric Vehicle EMS	Particle swarm optimization for hybrid electric vehicle energy management systems.	Suboptimal handling of rapid transitions in energy demand and supply.	Refine PSO algorithms to better manage rapid transitions and dynamic energy demands in hybrid systems.	[68]
State-of-Charge Estimation	Metaheuristic search methods for SOC estimation in Li-ion batteries.	High computational overhead in practical applications.	Develop more efficient and less computationally intensive algorithms for real-time SOC estimation.	[69]
Prediction of SOC for EVs	Prediction of SOC using various machine learning algorithms.	Lack of comprehensive validation across different EV models and battery types.	Perform validation across multiple EV models and battery types to ensure generalizability of results.	[70]
State-of-Charge Estimation in EV Batteries	Linear regression for SOC estimation in EV batteries.	Low accuracy in scenarios with rapid transitions in battery states.	Combine linear regression with other methods to enhance accuracy in rapidly changing conditions.	[71]
SOC and SOH Estimation for LIBs	Support vector machine for SOC and SOH estimation algorithms for lithium-ion batteries.	Insufficient testing across different LIB applications.	Expand testing across diverse LIB applications to improve generalization.	[72]
Fast Portable Charger for EVs	Sliding mode control for fast portable chargers for electric vehicles.	Limited performance data in real-world fast-charging scenarios.	Collect and analyze real-world performance data to enhance fast-charging algorithms.	[73]
Energy Prediction for EVs	Transformer neural network for energy prediction in electric vehicles.	Lacks robustness across different driving conditions.	Improve robustness of energy prediction models by incorporating a wider range of driving conditions.	[74]
Thermal Monitoring of LIBs	Fast recursive algorithm for thermal monitoring of lithium-ion batteries.	Limited to small-scale testing environments.	Expand thermal monitoring systems for large-scale and real-time applications in EVs.	[75]

summarizes research findings of AI optimization for EV smart charging systems.

Advanced AI models, such as Random Forest (RF), Recurrent Neural Network (RNN), Gaussian Process Regression (GPR), Support Vector Machine (SVM), and Reinforcement Learning (RL), provide substantial technological benefits in optimizing smart EV charging systems [45]. These models analyze historical and real-time data, such as user charging patterns and grid conditions, to make informed decisions that enhance charging efficiency and reduce operational costs. The station must also incorporate robust communication infrastructure to facilitate seamless data exchange between the charging station, electric vehicles, and the central management system. Advanced AI models for optimizing smart EV charging systems are shown in

Table 5. Advanced AI models for optimizing smart EV charging systems.

AI Model	Technological Benefits	Limitations	Solutions
Random Forest (RF)	Combines multiple decision trees for robust charging demand forecasting, reducing overfitting across varied datasets.	Can become computationally expensive with large datasets; may not handle real-time data well.	Use dimensionality reduction techniques and implement real-time data streaming solutions.
Recurrent Neural Network (RNN)	Handles sequential data, ideal for predicting time-series data such as user charging habits, enabling flexible load control.	Prone to vanishing gradient problems and may require extensive training time.	Implement Long Short-Term Memory (LSTM) networks or Gated Recurrent Units (GRUs) to address gradient issues.
Gaussian Process Regression (GPR)	Provides probabilistic regression for accurate demand forecasting with uncertainty quantification, aiding in risk-aware decision-making in charging operations.	Computationally intensive for large datasets; may struggle with high-dimensional data.	Utilize sparse approximations and dimensionality reduction techniques to improve scalability.
Support Vector Machine (SVM)	Optimizes margin-based classification for different charging scenarios, improving the precision of demand response strategies by better segregating load profiles.	May not perform well with very large datasets or when the number of features is high.	Use kernel approximations and scaling techniques to manage large feature sets and datasets.
Reinforcement Learning (RL)	Learns adaptive charging strategies through interaction with the environment, adjusting to grid conditions, price fluctuations, and user behavior.	Requires extensive training data and computational resources; may be sensitive to hyperparameter tuning.	Implement more efficient training algorithms and adaptive learning rate techniques to reduce training time and resource consumption.

.

In the realm of smart EV charging systems, integrating advanced AI optimization techniques such as load forecasting, state of charge (SOC) optimization, optimization of charging schedules, demand response (DR), renewable energy integration, energy storage systems (ESS), and tackling combined optimization problems are paramount. These elements collectively enable efficient management of electric vehicle charging at scale. Load forecasting ensures grid stability by predicting demand peaks and troughs, optimizing when and where EVs draw power [76]. SOC optimization improves battery efficiency and durability, which is essential for user happiness and sustainability. By adapting to real-time data, dynamic charging schedule optimization maximizes grid efficiency while minimizing expenses. Reducing dependency on fossil fuels through DR programs and the inclusion of renewable energy supports environmental objectives. By storing extra renewable energy and offering backup power, ESS increases resilience [77]. When combined, these AI-driven improvements guarantee that intelligent EV charging systems run smoothly, and consistently satisfy changing energy needs.

3. AI-Based Smart Charging Station

Advanced energy management techniques are integrated into AI-based smart charging stations to improve sustainability and efficiency. Demand response methods allow the system to dynamically modify charging schedules in response to user demand, energy pricing, and real-time grid circumstances [77]. This optimizes energy utilization and lessens the strain that peak load puts on the grid. Furthermore, the charging station may prioritize clean energy and lessen its need on non-renewable sources thanks to the incorporation of renewable energy sources like solar and wind. This strategy reduces the carbon footprint associated with charging electric vehicles while simultaneously fostering a more adaptable and robust energy infrastructure that can handle a range of energy supply and demand situations.

3.1. Mathematical Equations for AI-Based Smart Charging System

The key mathematical equations used in AI-based optimization for smart EV charging systems are given below.

Load forecasting is the process of predicting the total power demand from all electric vehicles at a given time. This is crucial for planning and ensuring that the power grid can handle the demand.

Load forecasting derivation is given below:

$$P_{EV}\left(t\right)={\sum }_{i=1}^N{P}_i\left(t\right)$$

(1)

where $P_{EV}\left(t\right)$ is total power demand from all EVs at time $t$, $N$ is the number of EVs, $P_i\left(t\right)$ is the power demand of the $i th$ EV at time $t$. The total power demand $P_{EV}\left(t\right)$ is calculated by summing the power demand $P_i\left(t\right)$ of all $N$ EVs at time $t$.

The state of charge (SOC) is a measure of the energy level in an EV battery. It is influenced by both the charging and discharging activities over time $\tau$. The SOC at any time $t$ is calculated by integrating the net power (charging power minus discharging power) over time, starting from an initial SOC, $SOC\left(0\right).$ This provides a continuous update of the battery’s energy level. The state of charge is calculated by

$$SOC(t)=SOC\left(0\right)+{\int }_0^t{\displaystyle \frac{P_{charge}\left(\tau \right)-P_{discharge}\left(\tau \right)}{C_{battery}}}d\left(\tau \right)$$

(2)

where $SOC\left(t\right)$ is the state of charge of the EV battery at time $t, SOC\left(0\right)$ is the initial state of charge. $P_{charge}\left(\tau \right)$ is the charging power at time $\tau$. $P_{discharge}\left(\tau \right)$ is the discharging power at time $\tau$, and $C_{battery}$ is the battery capacity.

Optimization of the charging schedule is formulated by

$$\min P_{charge}\left(t\right){\sum }_{t=1}^T(c\left(t\right).P_{charge}\left(t\right)+\lambda .{(P_{total}\left(t\right)-P_{target})}^2)$$

(3)

where the constraints are

$$P_{charge}\left(t\right)\le P_{max}$$

$${SOC}_{min}\le SOC\left(t\right)\le {SOC}_{max}$$

where $c\left(t\right)$ is the cost of electricity at time $t$. $\lambda$ is a penalty factor for deviation from the target. $P_{total}\left(t\right)$ is the total power demand at time $t$. $P_{target}$ is the target power demand. $P_{max}$ is the maximum charging power. ${SOC}_{min}$ and ${SOC}_{max}$ are the minimum and maximum state-of-charge limits. This optimization problem aims to minimize the cost of charging while ensuring that the power demand remains close to a target value and that the SOC is within acceptable limits. The objective function includes two terms: the cost of electricity (weighted by $c\left(t\right)$) and a penalty for deviating from the target power demand (weighted by $\lambda$). The constraints ensure that the charging power does not exceed a maximum limit and that the SOC stays within predefined bounds.

Demand response mechanisms adjust the charging power based on signals from the grid to maintain stability and efficiency [55]. The charging power $P_{charge}\left(t\right)$ is calculated by adding a baseline charging power $P_{base}\left(t\right)$ to a term that adjusts the power based on the difference between the grid signal $P_{signal}\left(t\right)$ and the target power $P_{target}$, scaled by the response coefficient $\alpha$.

Demand response is formulated by

$$P_{charge}\left(t\right)=P_{base}\left(t\right)+\alpha .(P_{signal}\left(t\right)-P_{target})$$

(4)

where $P_{base}\left(t\right)$ is the baseline charging power. $\alpha$ is the response coefficient indicating how much the charging power should adjust in response to grid signals. $P_{signal}\left(t\right)$ is the signal from the grid indicating the desired adjustment in power. $P_{target}$ is the target power demand.

Renewable energy integration involves using power generated from renewable sources like solar and wind to charge EVs. This is formulated by

$$P_{renewable}\left(t\right)=P_{solar}\left(t\right)+P_{wind}\left(t\right)$$

(5)

where the total renewable power $P_{renewable}\left(t\right)$ is the sum of the power generated from solar $P_{solar}\left(t\right)$ and wind $P_{wind}\left(t\right)$ sources at time $t$. Integration of renewable energy sources like solar and wind reduces reliance on the grid and enhances sustainability.

The net power required from the grid $P_{grid}\left(t\right)$ is the difference between the total power demand from EVs $P_{EV}\left(t\right)$ and the power generated from renewable sources $P_{renewable}\left(t\right)$. The formulation is given by

$$P_{grid}\left(t\right)=P_{EV}\left(t\right)-P_{renewable}\left(t\right)$$

(6)

Energy storage systems (ESS) play a crucial role in balancing supply and demand by storing excess energy and supplying it when needed. The net power flow $P_{ESS}\left(t\right)$ into or out of the ESS is calculated by considering the charging power $P_{charge}\left(t\right)$ multiplied by the charging efficiency ${\eta }_{charge}$ and the discharging power $P_{discharge}\left(t\right)$ divided by the discharging efficiency ${\eta }_{discharge}$, which is

$$P_{ESS}\left(t\right)={\eta }_{charge} . P_{charge}\left(t\right)-{\displaystyle \frac{P_{discharge}\left(t\right)}{{\eta }_{discharge}}}$$

(7)

This all-inclusive optimization issue seeks to minimize total cost while guaranteeing effective utilization of storage and renewable energy systems, preserving the targeted power levels, and adhering to SOC constraints. The cost of grid electricity and a penalty for deviating from the goal power demand are included in the objective function. The limitations guarantee that the SOC stays within predetermined boundaries, the charging power does not exceed the maximum limit, and the net power flow into or out of the ESS is appropriately modelled. The combined optimization problem is formulated by

$$\min P_{charge}\left(t\right),P_{discharge}\left(t\right) {\sum }_{t=1}^T(c\left(t\right).P_{grid}\left(t\right)+\lambda .{(P_{total}\left(t\right)-P_{target})}^2)$$

(8)

where the constraints are

$$P_{charge}\left(t\right)\le P_{max}$$

$${SOC}_{min}\le SOC\left(t\right)\le {SOC}_{max}$$

$$P_{grid}\left(t\right)=P_{EV}\left(t\right)-P_{renewable}\left(t\right)$$

$$P_{ESS}\left(t\right)={\eta }_{charge} . P_{charge}\left(t\right)-{\displaystyle \frac{P_{discharge}\left(t\right)}{{\eta }_{discharge}}}$$

The proposed mathematical formulas form the core of AI-driven optimization for intelligent electric car charging systems. The load forecasting, state of charge computation, charging schedule optimization, demand response, integration of renewable energy, and energy storage management are all covered by these equations. All of them work together to make electric car charging systems inexpensive, ecologically friendly, and efficient. Artificial intelligence is utilized to optimize various aspects of the system and achieve optimal results.

3.2. Implementation of an AI-Based Smart Charging Station

For example, an office complex has 10 electric vehicles (EVs) that need to be charged during the workday. The building has a total of 100 kWh of energy available for charging these EVs between 9 AM and 5 PM (8 h). The AI-based smart charging system optimizes the distribution of available power based on real-time demand, the state of charge (SOC) of each EV, and the time available.

Parameters:

• Total available energy: $E_{total}=100\mathrm{k}\mathrm{W}\mathrm{h}$
• Number of EVs: $N=10$
• Charging time: $T_{total}=8 \mathrm{h}$
• Initial state of charge for each EV: ${SOC\left(0\right)}_i$ (varies for each EV)
• Required SOC to fully charge an EV: ${SOC}_{full}=100\%$
• Power consumption rate per EV: $P_i=7 \mathrm{k}\mathrm{W}$ (assuming a standard Level 2 charger)

Step 1: Calculate Energy Required for Each EV

For each EVi, the energy required to reach full charge can be calculated as

$$E_i=({SOC}_{full}-{SOC\left(0\right)}_i)\times C_i$$

(9)

where $C_i$ is the battery capacity of EVi (in kWh).

Step 2: Distribute Power Using AI Optimization

The AI algorithm distributes the available power among the EVs based on

• Priority: EVs with lower SOC or higher urgency are prioritized.
• Efficiency: The system balances the load to avoid overloading the grid.

The power allocated to each EVi at time t is given by

$$P_i\left(t\right)={\displaystyle \frac{E_i}{T_{total}}}$$

(10)

The AI ensures the constraint: $\sum _{i=1}^N{P}_i\left(t\right)\le P_{max}$

Where $P_{max}$ is the maximum power capacity of the charging infrastructure.

Step 3: Calculate Energy Consumed Over Time

The total energy consumed by each EV over the charging period $T_{total}$ is

$$E_i^{consumed}=P_i\left(t\right)\times T_{total}$$

(11)

The AI algorithm adjusts $P_i\left(t\right)$ dynamically based on real-time data, ensuring that all vehicles receive sufficient charge within the available time and power constraints.

Calculation:

• EV 1 has an initial SOC of 20% and a battery capacity of 50 kWh.
• EV 2 has an initial SOC of 50% and a battery capacity of 60 kWh.

Energy required:

From Equation (9),

$$E_1=\left(1-0.20\right)\times 50=40 \mathrm{k}\mathrm{W}\mathrm{h}$$

$$E_2=\left(1-0.50\right)\times 60=30 \mathrm{k}\mathrm{W}\mathrm{h}$$

Power distribution:

From Equation (10),

$$P_1\left(t\right)={\displaystyle \frac{40}{8}}=5 \mathrm{k}\mathrm{W}$$

$$P_2\left(t\right)={\displaystyle \frac{30}{8}}=3.75 \mathrm{k}\mathrm{W}$$

Total energy consumed:

From Equation (11),

$$E_1^{consumed}=5\times 8=40 \mathrm{k}\mathrm{W}\mathrm{h}$$

$$E_2^{consumed}=3.75\times 8=30 \mathrm{k}\mathrm{W}\mathrm{h}$$

The AI algorithm adjusts $P_i\left(t\right)$ if other EVs require more energy or if constraints change.

3.3. Cost Comparison Analysis

Cost Components:

• Energy Cost ${(C}_E)$: The cost of electricity used to charge the EVs.

$$\mathrm{Traditional} \mathrm{System}: C_{E,T}=E_T\times {Rate}_{kWh}$$

(12)

where $E_T$ is the total energy consumed and ${Rate}_{kWh}$ is the cost per kWh.

$$\mathrm{AI}-\mathrm{Based} \mathrm{System}: C_{E,AI}=E_{AI}\times {Rate}_{kWh}$$

(13)

where $E_{AI}$ is the optimized energy consumption with the AI system.

• Operational Costs ${(C}_O)$: Costs related to the maintenance and operation of the charging infrastructure.

$$\mathrm{Traditional} \mathrm{System}: C_{O,T}=C_{maintanance,T}+C_{labor,T}$$

(14)

$$\mathrm{AI}-\mathrm{Based} \mathrm{System}: C_{O,AI}=C_{maintanance,AI}+C_{management,AI}$$

(15)

• Energy Savings $\left(S_E\right)$: Savings achieved by optimizing energy usage with the AI system.

$$S_E=\left({\displaystyle \frac{E_T-E_{AI}}{E_T}}\right)\times 100\%$$

(16)

• Total Cost Comparison: The total cost for each system is

$$\mathrm{Traditional} \mathrm{System}: C_{T,T}=C_{E,T}+C_{O,T}$$

(17)

$$\mathrm{AI} \mathrm{System}: C_{T,AI}=C_{E,AI}+C_{O,AI}$$

(18)

The percentage difference in total costs between the two systems is:

$${∆C}_T=\left({\displaystyle \frac{C_{T,T}-C_{T,AI}}{C_{T,T}}}\right)\times 100\%$$

(19)

Calculation:

• Traditional System: Consumes the full 100 kWh.
• AI-Based System: Optimized to consume only 70 kWh.
• Rate per kWh: ${Rate}_{kWh}$ = 0.20 USD.
• Operational Costs:

Traditional: $C_{O,T}$ = 500 USD.

AI-Based: $C_{O,AI}$ = 400 USD (considering maintenance and AI management).

Step 1: Calculate Energy Costs

From Equation (12),

$$\mathrm{Traditional} \mathrm{System} C_{E,T}=100\times 0.20=20 \mathrm{USD}$$

From Equation (13),

$$\mathrm{AI}-\mathrm{Based} \mathrm{System}: C_{E,AI}=70\times 0.20=14 \mathrm{USD}$$

Step 2: Calculate Total Costs

From Equation (17),

$$\mathrm{Traditional} \mathrm{System}: C_{T,T}=20+500=520 \mathrm{USD}$$

From Equation (18),

$$\mathrm{AI}-\mathrm{Based} \mathrm{System}: C_{T,AI}=14+400=414 \mathrm{USD}$$

Step 3: Calculate Energy Savings

From Equation (16),

$$S_E=\left({\displaystyle \frac{100-70}{100}}\right)\times 100\%=30\%$$

Step 4: Calculate Total Cost Difference

From Equation (19),

$${∆C}_T=\left({\displaystyle \frac{520-414}{520}}\right)\times 100\%=20.38\%$$

These percentages indicate the relative efficiency and cost-effectiveness of the AI-based system over the traditional charging method. The AI-based system not only saves energy but also reduces overall operational costs, providing significant savings in the long run.

4. AI-Driven Innovations in Malaysian EV Charging Systems

In order to facilitate the growing national adoption of electric vehicles, Malaysia is actively building up its infrastructure for EV charging [78]. In order to accommodate the rising number of EV users, the nation has been developing its public charging network, concentrating on urban areas and main routes. The government has put in place programs and incentives to promote the adoption of EVs, such as tax breaks for EV purchases and funding for the construction of charging infrastructure. Figure 11 shows total vehicle industry volume with variance. There is an overall increase of 6.6% in the total number of vehicles from 1H2023 to 1H2024, equating to an additional 24,120 units [10]. Passenger vehicles saw a notable growth of 9.2%, with an increase of 30,162 units from 1H2023 to 1H2024. This suggests a significant rise in consumer demand for passenger vehicles. Interestingly, commercial vehicles experienced a decrease of 15.3%, translating to 6042 fewer units in 1H2024 compared to 1H2023. This decline might reflect changing market conditions, reduced demand, or shifts in commercial transportation strategies [10].

To ensure compatibility with various EV models, Malaysia is progressively implementing the CCS (Combined Charging System) for DC fast charging, while still relying mostly on Type 2 connections for AC charging. Even with these developments, there are still issues such the requirement for additional rural charging stations and maintaining the dependability and upkeep of the current infrastructure. In the future, Malaysia wants to further its objectives of low carbon emissions and sustainable transportation by integrating renewable energy sources, improving charging speeds, and growing its network of electric vehicle charging stations.

Table 6. EV charging infrastructure providers and networks.

Provider	Network Name	Locations	Charger Types
BMW Malaysia	BMW i Charging	BMW dealerships, selected locations	Mode 3 AC, Mode 4 DC
Chargehere EV Solutions	Chargesini	Nationwide	Mode 3 AC, Mode 4 DC
DC handal	DC handal	Various location including offices, malls	Mode 4 DC
EV Connection	JomCharge	Nationwide	Mode 3 AC, Mode 4 DC
Petronas	Gentari	Nationwide	Mode 3 AC, Mode 4 DC
Shell Malaysia	Shell Recharge	Nationwide	Mode 3 AC, Mode 4 DC
Tesla	Tesla	Various location including malls, offices, etc.	Mode 3 AC, Mode 4 DC
TNB (Tenaga Nasional Berhad)	TNB Electron	Selected locations as part of pilot projects	Mode 4 DC
Yinson Greentech Malaysia	ChargeEV	Nationwide	Mode 3 AC, Mode 4 DC

lists Malaysia’s networks and providers of EV charging infrastructure.

Several nations, including the US, China, and several in Europe, have made significant strides in building extensive EV charging networks [44]. Wireless charging, AI-powered smart charging and V2G technologies have raised industry standards. Studying other countries’ successful AI implementations in EV charging networks is instructive. European programs like GridMotion study V2G technology to stabilize grids, while Tesla’s Supercharger network optimizes charging times and distribution via AI. Artificial intelligence-driven solutions are revolutionizing Malaysia’s EV charging environment by improving efficiency, reliability, and user experience. These innovations increase grid integration, dynamic charging station management, and demand forecasting using AI. AI algorithms assess real-time traffic, weather, and user behavior data to maximize charging schedules and energy distribution. This predictive capability balances load and integrates renewable sources effectively, promoting sustainable energy and operational efficiency. AI-driven solutions also personalize consumer experiences with intelligent suggestion systems, immediate charging station availability information, and simple payment options. Malaysia is leading the way in the development of a more intelligent and responsive EV charging infrastructure that is in line with global trends towards sustainable mobility and smart city projects by encouraging cooperation between technology suppliers, governmental organizations, and utilities [79]. Malaysia’s approach to smart EV charging compared to other countries is shown in

Table 7. Malaysia’s approach to smart EV charging.

Aspect	Malaysia	Other Countries (e.g., USA, Germany, China)
Regulatory Framework	Development of EV policies and incentives	Stringency of regulations, incentives for EV adoption
Infrastructure Deployment	Coverage and density of charging stations	Extent of public and private charging infrastructure
Technological Integration	Adoption of AI, IoT in charging networks	Use of advanced technologies in grid management and analytics
Public–Private Partnerships	Collaboration between government and industry	Partnership models supporting rapid infrastructure development
User Experience	Reduce range anxiety; increase accessibility and payment options	Convenience, reliability, and user feedback mechanisms
Environmental Sustainability	Integration with renewable energy sources	Green energy initiatives, carbon footprint reduction strategies
Innovation Ecosystem	Support for startups and tech innovation	Investment in R&D, incubation programs

.

As of 2022, the Klang Valley is home to roughly 9 million people in Malaysia. There are 263,220 serviced apartment units in Malaysia [80]. Assuming an average of two cars per serviced apartment unit in Malaysia and 30% of BEVs, as targeted by the LCMB, there will be a total of 157,932 BEVs by 2030. Some high-rise building managers do not allow EV users to install charging stations in the building, even if the users are paying for it. Sometimes, this is because the building’s power supply distribution board at the car park does not have the power margin for EV chargers [81]. Existing building electrical capacity and also EV driver user demand will be two parameters of importance in determining the demand response control mechanisms. Figure 12 shows energy usage by condominium in each Malaysian state.

The installation of a 47 kW DC charger (Dual Gun CCS2) × 4 nos at The Curve in Mutiara Damansara, situated alongside the LDP (Malay: Lebuhraya Damansara–Puchong) Highway, represents a significant advancement in Malaysia’s electric vehicle (EV) charging infrastructure, facilitated by Chargehere EV Solution Sdn. Bhd. (Platform: Chargesini); it is shown in Figure 13. Tesla’s largest charging station in Southeast Asia (2024) in Gamuda Cove—a smart and sustainable low-carbon city, Sepang, Selangor Darul Ehsan, Malaysia—is shown in Figure 14. This DC charger not only accommodates rapid charging for EVs but also integrates sophisticated technologies for monitoring and data collection. Robust mechanisms are incorporated into Chargehere EV Solution Sdn. Bhd. (Platform: Chargesini)’s implementation to gather cloud data and charging parameters from EV chargers. Important parameters like voltage, current, power output (kW), and charging time must be monitored in real time for this to occur. These variables are necessary to guarantee effective energy transmission, maximize battery life, and optimize charge durations.

The ability to gather data in the cloud makes it possible to effortlessly incorporate charging data into a centralized platform that is accessible from a distance [82]. Operators may track charger performance, anticipate maintenance requirements, and examine consumption trends by utilizing cloud-based information. This is vital information to track and analyze EV driver behavior that allows AI algorithm prediction training; with this information on board, the AI-driven smart charging system allows optimization of electrical infrastructure utilization and enhances user experience, especially in reducing range anxiety concerns. This also balances energy demand and speeds up the switch to renewable energy sources while also improving operational efficiency and supporting smart grid integration initiatives.

Further improving EV users’ comfort while travelling along the LDP Highway is this installation at The Curve, which offers dependable access to fast charging stations. It highlights Malaysia’s dedication to environmentally friendly transportation options and establishes Chargehere EV Solution Sdn. Bhd. (Platform: Chargesini) as a pioneer in the development of EV infrastructure. All things considered, the integration of a DC charger with extensive charging parameter monitoring and cloud data collecting demonstrates how contemporary technologies are propelling the development of EV charging networks and enhancing Malaysia’s transportation ecosystem in a cleaner and more effective manner. Challenges and solutions in implementing smart charging systems in Malaysia are shown in

Table 8. Challenges and solutions in implementing smart charging systems in Malaysia.

Aspect	Challenges	Solutions
Technical Compatibility	Interoperability between EVs and charging infrastructure, diverse communication protocols, and integration with smart grid systems.	Standardize connector types (e.g., CCS, CHAdeMO), develop universal communication protocols, and integrate advanced grid management technologies.
Regulatory Framework	Developing standards, tariff structures, and regulatory policies supporting smart charging deployment.	Establish clear regulatory guidelines, incentivize compliance with subsidies, and collaborate with stakeholders to streamline regulatory processes.
Financial Viability	High initial setup costs, ongoing maintenance expenses, and securing funding sources for infrastructure deployment and operation.	Seek government grants, private investments, and develop sustainable financial models with cost-effective maintenance strategies.
Operational Efficiency	Ensuring data privacy and security, optimizing scalability, overcoming user acceptance barriers, and integrating with existing energy management systems.	Implement robust cybersecurity protocols, educate users on benefits, design scalable systems with modular architecture, and integrate seamlessly with existing infrastructure.
Environmental Impact	Managing energy demand, optimizing energy use from renewable sources, and reducing carbon footprint through smart charging strategies.	Maximize use of renewable energy sources, optimize charging schedules based on energy demand and availability, and implement energy-efficient charging practices.
User Experience	Educating users on benefits, ensuring ease of use, and addressing user concerns such as range anxiety and charging availability.	Provide user-friendly interfaces and real-time updates on charging station availability, and educate users on the benefits of smart charging and EV ownership.
Technological Innovation	Integrating AI, IoT, and advanced analytics for predictive maintenance, dynamic energy management, and enhanced grid interaction.	Invest in AI-driven solutions for predictive maintenance, dynamic energy management, and grid interaction optimization, fostering continuous technological advancement.
Grid Limitations	Grid capacity constraints during peak demand periods, potential grid instability due to simultaneous EV charging, and strain on local distribution networks.	AI-driven systems that optimize charging schedules based on real-time demand, prioritizing off-peak charging to alleviate peak demand strain. Enable EVs to send power back to the grid during peak demand, helping stabilize the grid and act as distributed energy resources.

.

Some companies and initiatives represent key players and efforts in Malaysia’s evolving EV and smart charging landscape. They contribute to infrastructure development, technology integration, and policy advocacy aimed at fostering sustainable mobility solutions across the country.

AI-driven solutions are revolutionizing Malaysia’s EV charging infrastructure by improving efficiency, reliability, and user engagement. By using cutting-edge AI algorithms, Malaysia may increase grid integration to support renewable energy sources, better anticipate and manage demand, and improve charging infrastructure efficiency. These innovations simplify charging and enable real-time updates and intelligent suggestions to personalize the user experience. AI-driven electric vehicle (EV) charging advances are expected to be crucial in creating a more environmentally friendly and technologically advanced transportation ecosystem, especially as Malaysia invests in smart city projects and sustainable mobility solutions.

5. Future Prospects and Recommendations

AI might change sustainable transportation in Malaysia’s smart EV charging networks [83]. AI can increase EV charging infrastructure efficiency, reliability, and user experience. AI-driven predictive analytics may improve charging station operations by anticipating demand trends, regulating grid load, and integrating renewable energy sources [84,85]. This predictive capacity improves energy management and helps Malaysia reduce carbon emissions and promote green technology [86]. AI-driven predictive maintenance may also detect and fix equipment issues, reducing charging station downtime. Proactive maintenance reduces operational costs, boosts customer satisfaction, and ensures service reliability [87]. AI also improves EV charging station user experience. AI-derived ideas can improve convenience and accessibility for EV customers by providing real-time information on charging station availability, energy cost-optimal charging periods, and nearby amenities.

Several suggestions can maximize AI’s potential in Malaysia’s EV charging infrastructure:

i.

Incompatibility with EV Models: Currently, not all EV models can handle every charging level, and not all public charging stations are equipped for high-power devices. This discrepancy creates challenges for EV owners in finding suitable charging infrastructure [88].

ii.

Fixed Demand Fees: Users of fast charging stations often face a fixed monthly demand fee, which can deter EV owners from using these facilities due to the inability to charge based on a variable power tariff. Revising the fixed demand fee policy could address some of these concerns [89].

iii.

Inconsistent Charging Facility Layouts: Charging station layouts vary significantly because they are installed by different businesses in diverse locations. Users struggle with these inconsistencies, and a standardized charging facility layout, similar to conventional internal combustion engine vehicle refueling stations, could enhance the user experience [90].

iv.

Barriers to Private Fast-Charging Facilities: Establishing private fast-charging facilities, such as those in homes, is often challenging and requires permissions from local service providers and government authorities. This cumbersome process can discourage EV owners from setting up their own private charging infrastructure.

v.

Strategic Deployment: There is a need for strategic placement of EV charging stations along major roads and in urban areas. The current lack of planning for charging stations outside major cities is a concern for EV owners [91].

vi.

Integration of Renewable Energy: Charging stations powered by renewable energy sources, such as solar or wind, require substantial space and investment for design and implementation. Vacant lots near roadways present promising locations for these renewable energy-powered EV charging stations [92,93].

vii.

Investment in Research and Development: Promote cooperation between the public and private sectors to advance AI innovation in EV charging technology. Research funds and incentives should be given out to encourage the creation of AI-driven solutions that are suited to regional needs.

viii.

Standardization and Interoperability: To guarantee a smooth user experience and uninterrupted access across many platforms and places, establish guidelines for data exchange and interoperability among charging networks.

ix.

Public Awareness and Education: Inform customers and interested parties about the advantages of AI-powered smart charging systems while addressing issues with dependability, security, and data privacy.

x.

Incentives for AI Adoption: Offer financial incentives, tax breaks, or subsidies to businesses investing in AI technologies for EV infrastructure, accelerating deployment and scalability [94].

xi.

Regulatory Framework: Provide flexible regulatory frameworks that promote AI technical developments and guarantee sustainability, equity, and safety in EV charging operations [95]. Encourage the creation of legal frameworks that support innovation and make it easier to integrate AI technologies into the infrastructure that is already in place.

By implementing these suggestions, Malaysia can become a leader in intelligent transport solutions, boost economic growth through technical innovation, and contribute to global efforts to reduce carbon emissions and promote sustainable mobility. AI-driven smart EV charging systems have the potential to revolutionize transportation and create a greener, more efficient, and user-centric future.

6. Conclusions

An important step towards a more efficient, sustainable, and user-friendly transportation infrastructure has been taken by Malaysia with the integration of artificial intelligence (AI) into their smart electric vehicle charging systems. With the help of AI, energy management can be optimized, demand patterns can be predicted, and EV charging stations can be made more operationally reliable. The use of AI-driven predictive analytics will help Malaysia achieve its goals of reducing carbon emissions and boosting environmental sustainability by proactively managing grid integration, minimizing impacts of peak loads, and encouraging the integration of renewable energy sources.

In addition to reducing downtime and increasing service availability, AI-enabled predictive maintenance increases the lifetime and resilience of charging infrastructure. This forward-thinking approach boosts operational efficiency and delights customers with simple and reliable charging alternatives. The usefulness of artificial intelligence in electric vehicle charging systems is illustrated by the enhanced user experiences that result from customized recommendations and real-time data on the availability of charging stations. Malaysia can encourage more people to use electric cars and speed up the shift to cleaner transportation options by making charging stations more conveniently located and by adjusting charging schedules according to projected energy prices and demand.

To ensure future innovation in electric vehicle charging operations that are sustainable, equitable, and safe, Malaysia should prioritize investing in artificial intelligence research and development, promote public–private partnerships, and construct robust legislative frameworks. Public education initiatives and financial incentives for artificial intelligence usage will substantially facilitate the widespread implementation of smart electric vehicle charging infrastructure. An important opportunity for Malaysia to become a global leader in smart mobility solutions has presented itself in the form of smart EV charging systems powered by artificial intelligence. Incorporating these ideas and technology into Malaysia’s infrastructure would allow the country to build a strong system that can handle its current and future energy demands while also paving the way for a greener, more sustainable transportation system.

The scalable deployment of smart electric vehicle charging systems in Malaysia will be the subject of future research that aims to improve AI algorithms for grid optimization and demand forecasting, integrate edge computing for real-time monitoring, provide personalized services to enhance user experience, tackle cybersecurity issues, build collaborative environments, develop supportive policy frameworks, and analyze socioeconomic impacts.