Artificial Intelligence-driven control of lithium-ion battery systems for improved grid integration and power supply optimization

 Corresponding author: Paul Uchechukwu Nzereogu; Email: [email protected] Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0. Artificial Intelligence-driven control of lithium-ion battery systems for improved grid integration and power supply optimization Paul Uchechukwu Nzereogu 1, *, Adam Oluwadamilola Yoonus 2, Enoch Temiloluwa Olonade 3, Ugochukwu Humphrey Ibekwe 4, Oluwafemi Odunayo Olusesan 5, Ibrahim Lawal Abdullahi 6 and Victor Ikechukwu Stephen 7 1 Department of Metallurgical and Materials Engineering, University of Nigeria, Nsukka. 2 Department of Materials and Metallurgical Engineering, University of Ilorin, Ilorin, Nigeria. 3 Department of Mechanical Engineering, University of Ilorin, Ilorin, Nigeria. 4 Department of Chemical Engineering, Federal University of Technology Owerri, Nigeria. 5 Department of Materials and Metallurgical Engineering, University of Ilorin, Ilorin, Nigeria. 6 Department of Physics, Tezpur University, Assam 784028, India. 7 Department of Electrical and Electronics Engineering, Michael Okpara University of Agriculture, Umudike, Nigeria. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 Publication history: Received on 02 August 2025; revised on 11 September 2025; accepted on 13 September 2025 Article DOI: https://doi.org/10.30574/gjeta.2025.24.3.0263 Abstract The rapid proliferation of renewable energy sources has underscored the critical need for robust energy storage solutions to ensure grid stability and efficient power delivery. Lithium-ion battery (LIB) systems stand at the forefront of this transition, yet their performance in grid-scale applications is often hampered by challenges such as degradation, thermal instabilities, and suboptimal integration. This review explores the transformative potential of artificial intelligence (AI)-driven control strategies in overcoming these barriers, synthesizing recent advancements in machine learning, deep learning, and reinforcement learning for battery management systems (BMS). By examining AI applications in state estimation, thermal management, grid stability, and power supply optimization, the paper highlights how these technologies enable precise energy dispatch, enhance demand response, and facilitate seamless coordination with smart grids. Key insights reveal significant improvements in LIB efficiency, lifespan extension, and economic viability, while also yielding environmental benefits through reduced carbon emissions and minimized renewable curtailment. Ultimately, this work illuminates the path toward resilient, sustainable power infrastructures, identifying persistent challenges like data dependency and computational demands, alongside promising avenues for innovation. As AI continues to evolve, its integration with LIB systems promises to redefine grid dynamics, offering a compelling blueprint for future energy systems that balance reliability, cost, and ecological imperatives. Keywords: Lithium-Ion Batteries; Artificial Intelligence; Battery Management Systems; Grid Integration; Power Supply Optimization; Renewable Energy; Smart Grids; Energy Storage 1. Introduction The escalating demand for sustainable energy solutions amid climate change concerns has propelled lithium-ion battery (LIB) systems to the forefront of modern power infrastructures. These batteries serve as pivotal components in bridging intermittent renewable energy sources with reliable grid operations, enabling efficient storage and dispatch of power. Artificial intelligence (AI)-driven control mechanisms have emerged as transformative tools to address inherent limitations in LIB performance, such as degradation and thermal instabilities, thereby optimizing grid integration and Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 282 power supply. This review synthesizes recent advancements in AI applications for LIB management, highlighting their potential to enhance system reliability and economic viability in energy ecosystems. 1.1. Background and Importance of Lithium-Ion Battery Systems Lithium-ion batteries operate on the principle of reversible intercalation of lithium ions between positive and negative electrodes, typically composed of lithium metal oxides for the cathode and graphite or silicon-based materials for the anode, facilitated by an electrolyte that enables ion transport. This electrochemical process yields high energy densities, often exceeding 250 Wh/kg, and efficiencies above 95%, making LIBs superior to alternatives like lead-acid or nickelmetal hydride batteries in terms of cycle life and power output. The scalability of LIB systems allows for deployment in various capacities, from portable devices to megawatt-scale installations, underscoring their versatility in energy storage applications. As global energy demands surge, LIBs have become indispensable for buffering fluctuations in power supply, particularly in scenarios involving high penetration of renewables. Stroe et al. [1] emphasized the suitability of specific LIB chemistries, such as LiMn2O4/Li4Ti5O12 and LiFePO4/C, for grid services like primary frequency regulation, demonstrating through modeling that these cells maintain performance under cyclic stress from wind power integration. The integration of LIBs with renewable energy systems, such as solar photovoltaic and wind farms, addresses the intermittency challenge by storing excess energy during peak generation and releasing it during deficits, thereby stabilizing grid voltage and frequency. In hybrid setups, LIBs act as buffers that smooth power output, reducing the need for fossil fuel-based peaking plants and lowering overall carbon emissions. For instance, in microgrid configurations, LIBs enable islanded operations where renewable sources dominate, ensuring uninterrupted supply to remote or critical loads. Khalid et al. [2] provided an overview of technical specifications for battery energy storage systems (BESS) in microgrids, noting that LIBs offer rapid response times—often in milliseconds—for ancillary services like load shifting and voltage support, which are crucial for maintaining grid resilience amid variable renewable inputs. Furthermore, the economic importance of LIBs lies in their declining costs, driven by advancements in manufacturing, which have dropped below $100/kWh in recent years, making large-scale deployment feasible for utilities aiming to achieve net-zero targets. Despite these advantages, grid-scale applications of LIBs face significant challenges, including capacity fade due to calendar aging and cycling, which can reduce usable energy by up to 20% over 5-10 years. Thermal runaway risks, exacerbated by overcharging or physical damage, pose safety hazards that necessitate robust monitoring. Additionally, inconsistencies in cell manufacturing led to imbalances within battery packs, affecting overall efficiency and lifespan. Uddin et al. [3] investigated the impact of high-frequency current ripple on LIB performance in electric vehicles, finding that such ripples accelerate degradation through increased internal resistance and heat generation, a concern that extends to grid applications where variable loads mimic similar stresses. These issues highlight the need for advanced management strategies to mitigate losses and extend operational life, particularly in high-demand grid environments. The broader importance of LIB systems extends to enabling vehicle-to-grid (V2G) interactions, where electric vehicle batteries contribute to grid balancing by supplying power during peaks. This bidirectional flow not only optimizes power supply but also generates revenue for users through energy arbitrage. In utility-scale contexts, LIBs support black-start capabilities and blackouts prevention, enhancing energy security. Hakam et al. [4] explored intelligent V2G and vehicle-for-grid (V4G) systems using artificial neural networks, illustrating how LIBs facilitate smart grid functionalities by predicting and adapting to demand patterns, thus underscoring their role in future-proofing power infrastructures against volatility. 1.2. Role of Artificial Intelligence in Battery Management Artificial intelligence encompasses a suite of computational techniques, including machine learning (ML), deep learning (DL), and reinforcement learning (RL), that enable predictive modeling and adaptive control in battery management systems (BMS). In LIB contexts, AI processes vast datasets from sensors monitoring voltage, current, and temperature to optimize operations beyond traditional rule-based methods. By learning from historical patterns, AI algorithms forecast battery behavior under diverse conditions, reducing downtime and improving efficiency. Lipu et al. [5] conducted a statistical analysis of AI approaches in electric vehicle BMS, revealing that ML techniques enhance state estimation accuracy by up to 30%, paving the way for proactive maintenance and extended battery life in gridintegrated setups. In state estimation, AI plays a crucial role in determining state of charge (SOC) and state of health (SOH), parameters essential for preventing over-discharge or overcharge that could compromise grid reliability. Neural networks, for example, integrate Kalman filters with data-driven models to handle nonlinearities in battery dynamics, achieving Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 283 estimation errors below 1%. This precision is vital for power supply optimization, where accurate SOC informs dispatch decisions in BESS. Boujoudar et al. [6] demonstrated the use of artificial neural networks for LIB modeling and SOC estimation, showing that AI outperforms equivalent circuit models in capturing electrochemical nuances, particularly under variable grid loads. Thermal management and safety are further bolstered by AI, which employs predictive analytics to detect anomalies like hotspots before they escalate into failures. DL models analyze thermal imaging and sensor data to regulate cooling systems dynamically, minimizing energy waste while ensuring operational safety. In grid applications, this prevents cascading failures that could disrupt power supply. Wang et al. [7] applied surrogate-based multidisciplinary optimization using AI for LIB thermal management in electric vehicles, finding that AI-driven designs reduce temperature variances by 15%, a benefit translatable to grid-scale systems where thermal stability directly impacts integration efficiency. Overall, AI fosters grid integration by enabling real-time optimization of charge/discharge cycles, aligning battery operations with grid demands for peak shaving and frequency regulation. Through RL, systems learn optimal policies for energy arbitrage, maximizing economic returns. Wali et al. [8] reviewed grid-connected LIB BESS, highlighting how AI integration in bibliometric analyses points to trends in predictive control, which enhance power supply reliability and support renewable-heavy grids. Objectives and Scope of the Review The primary objective of this review is to critically evaluate AI-driven control strategies for LIB systems, with a focus on improving grid integration and power supply optimization. By synthesizing literature on AI applications, the paper aims to identify effective techniques for enhancing battery performance metrics like efficiency and lifespan in dynamic grid environments. This includes assessing how AI mitigates challenges such as degradation and intermittency, ultimately contributing to sustainable energy frameworks. Oyucu et al. [9] optimized LIB performance using ML and explainable AI, illustrating objectives aligned with energy management enhancements that inform grid-scale adaptations. The scope encompasses key AI methodologies—ML, DL, and RL—applied to BMS functions like state prediction, fault diagnosis, and energy dispatch, while excluding non-LIB technologies or purely theoretical models without empirical validation. Emphasis is placed on grid-specific applications, including V2G and microgrids, drawing from studies post2014 to capture recent advancements. This delimited focus ensures relevance to current challenges in power optimization. Gjelaj et al. [10] examined grid integration of fast-charging stations using modular LIBs, defining a scope that parallels this review's emphasis on AI-enhanced scalability and efficiency. The structure of the review progresses from LIB fundamentals to AI techniques, grid integration, optimization strategies, challenges, and future directions, providing a logical flow for readers. Each section incorporates case studies and comparative analyses to substantiate claims. This organization facilitates a comprehensive understanding of AI's transformative potential. Finally, the review contributes by highlighting research gaps, such as AI scalability in large grids, and proposes directions for hybrid AI-physical models. Its significance lies in guiding policymakers and engineers toward AI implementations that accelerate the transition to optimized, resilient power systems. 2. Fundamentals of Lithium-Ion Battery Systems 2.1. Operating Principles and Characteristics Lithium-ion batteries (LIBs) function through the reversible movement of lithium ions between a cathode, typically composed of lithium metal oxides like LiCoO2 or LiFePO4, and an anode, often graphite or silicon-based, mediated by a liquid or solid electrolyte. Figure 1 illustrates the basic schematic of a lithium-ion battery, highlighting the key components and ion flow that underpin its high energy density and efficiency. During discharge, lithium ions migrate from the anode to the cathode through the electrolyte, generating an electric current via electron flow in an external circuit, with the reverse occurring during charging. This electrochemical process yields high energy densities (150–250 Wh/kg) and power density, alongside efficiencies exceeding 95%, positioning LIBs as the preferred choice for energy storage in both mobile and stationary applications. According to Goodenough and Kim [11], the development of layered oxide cathodes and carbon-based anodes has significantly enhanced LIB performance, enabling high voltage outputs (3.6–3.8 V nominal) and long cycle lives, often exceeding 1000 cycles under optimal conditions. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 284 The performance metrics of LIBs, such as capacity, cycle life, and efficiency, depend heavily on material selection and cell design. For instance, LiFePO4 cathodes offer superior thermal stability and safety compared to LiCoO2, albeit at a lower energy density, making them suitable for grid-scale applications where safety is paramount. The electrolyte, typically a lithium salt like LiPF6 dissolved in organic solvents, facilitates ion transport but is sensitive to temperature and overvoltage, which can trigger side reactions and capacity fade. A study by Kalhoff et al. [12] highlighted how advancements in electrolyte additives, such as vinylene carbonate, improve the formation of a stable solid electrolyte interphase (SEI) layer, reducing capacity loss by up to 15% over 500 cycles, a critical factor for grid applications requiring consistent performance. Figure 1 Schematic representation of a lithium-ion battery (showing the cathode, anode, electrolyte, and ion movement during charge/discharge). Reproduced with permission from Ref [11], under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/) LIBs are distinguished by their scalability and modularity, enabling configurations from small consumer electronics to megawatt-hour-scale battery energy storage systems (BESS). Their fast response time, often in milliseconds, makes them ideal for applications requiring rapid power delivery, such as frequency regulation in power grids. However, internal resistance and polarization effects can limit performance under high current demands, necessitating precise control systems to optimize operation. Findings from Aurbach et al. [13] indicate that optimizing electrode porosity and thickness can mitigate these limitations, enhancing power output by 10–20% in high-rate applications, which is essential for grid integration where dynamic load responses are common. The versatility of LIBs extends to their ability to operate across a wide temperature range (-20°C to 60°C), though performance degrades outside optimal conditions (15–35°C). This adaptability, coupled with declining costs (below $100/kWh in 2023), has driven widespread adoption in renewable energy systems. Yet, the choice of chemistry impacts performance trade-offs; for example, nickel-rich cathodes (LiNiMnCoO2) offer higher capacity but reduced thermal stability compared to LiFePO4. These characteristics underscore the need for tailored management strategies to maximize LIB utility in grid contexts. To provide a comprehensive overview of LIB variations suitable for grid integration, Table 1 compares key chemistries based on performance metrics, advantages, and drawbacks, drawing from recent studies. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 285 Table 1 Comparison of Lithium-Ion Battery Chemistries for Grid Applications Chemistry Cathode Material Anode Material Energy Density (Who/kg) Cycle Life (Cycles) Thermal Stability Cost ($/kWh) Safety Rating (1-5, 5=Best) Key Advantages Key Disadvantages References LiCoO2 (LCO) Lithium Cobalt Oxide Graphite 150-200 5001000 Low 150-200 2 High energy density, mature technology Poor thermal stability, expensive cobalt [11], [12] LiFePO4 (LFP) Lithium Iron Phosphate Graphite 90-160 20005000 High 100-150 5 Excellent safety, long cycle life, stable Lower energy density, slower charge rates [11], [14] LiMn2O4 (LMO) Lithium Manganese Oxide Graphite/Li4Ti5O12 100-150 10002000 Medium 120-180 3 Good thermal stability, low cost Moderate cycle life, manganese dissolution [1], [13] LiNiMnCoO2 (NMC) Lithium Nickel Manganese Cobalt Oxide Graphite/Silicon 200-250 10003000 Medium 100-150 3 Balanced performance, high capacity Nickel-rich variants less stable, cobalt dependency [11], [15] LiNiCoAlO2 (NCA) Lithium Nickel Cobalt Aluminum Oxide Graphite 200-260 10002000 Medium 120-170 3 High energy, good power output Thermal runaway risk, aluminum sensitivity [14], [15] Li4Ti5O12 (LTO) Lithium Titanate (as anode with various cathodes) Lithium Titanate 70-100 500010000 High 200-300 5 Ultra-long life, fast charging, safe Low energy density, higher cost [1], [12] Silicon-based Various (e.g., NMC) Silicon-Graphite Composite 250-350 5001500 Low 150-250 2 Very high capacity, potential for density gains Volume expansion, rapid degradation [13], [14] Solid-State (Emerging) Lithium Metal Oxides Lithium Metal/Graphite 300-500 10003000 High 200-400 4 Improved safety, higher density Scalability issues, high manufacturing cost [12], [18] Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 286 2.2. Challenges in Grid-Scale Applications The deployment of LIBs in grid-scale applications introduces challenges related to longevity, safety, and operational efficiency. Capacity fade, resulting from calendar aging and cycling, reduces available energy over time, with some cells losing 20–30% of capacity within 5–10 years under typical grid conditions. As depicted in Figure 2, capacity fade accelerates under varying DODs and temperatures, emphasizing the need for advanced mitigation strategies in gridscale deployments. This degradation stems from SEI growth, lithium plating, and electrode material breakdown, particularly under high charge/discharge rates. According to Luo et al. [14], accelerated aging tests reveal that high Crate cycling (e.g., 2C) can double capacity fade rates compared to low-rate cycling (0.5C), posing a significant barrier for BESS tasked with frequent power balancing in grids with high renewable penetration. Thermal runaway remains a critical safety concern, triggered by overcharging, short circuits, or mechanical damage, which can lead to fires or explosions in large-scale installations. The risk is amplified in grid applications where thousands of cells are interconnected, as a single cell failure can propagate through the pack. In a comprehensive study, Feng et al. [15] analyzed thermal runaway mechanisms, noting that gas generation from electrolyte decomposition at elevated temperatures (>150°C) initiates chain reactions, necessitating advanced cooling and monitoring systems to ensure safety. These findings emphasize the importance of real-time diagnostics to prevent catastrophic failures in gridscale BESS. Figure 2 Capacity fade curves under different depths of discharge (DODs) and temperatures: (a) DOD = 30%. (b) DOD = 40%. (c) DOD = 50%. (d) DOD = 60%. (e) Capacity loss per cycle before the ‘knee point’ with different DODs and mean SOC values. Reproduced with permission from Ref [14], under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/) Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 287 Table 2 Major Challenges in Grid-Scale LIB Applications and Mitigation Strategies Challenge Category Specific Issue Impact on Grid Integration Severity (Low/ Medium/ High) Common Causes Mitigation Strategies Potential Improvement (%) Examples from Literature References Degradation Capacity Fade Reduced usable energy (20-30% loss over 5-10 years) High SEI growth, lithium plating, cycling stress AI-based predictive maintenance, optimal charging protocols 15-25 Accelerated aging tests at high C-rates [3], [14] Thermal Instabilities Thermal Runaway Fire/explosion risks, system failures High Overcharging, short circuits, high temperatures (>150°C) Dynamic cooling systems, AI anomaly detection 12-20 Gas generation from electrolyte decomposition [7], [15] Cell Imbalances Voltage/Capacity Variations Reduced pack efficiency (15% performance drop) Medium Manufacturing tolerances, uneven aging Active balancing algorithms, MLbased monitoring 10-15 5% initial capacity variation effects [16], [21] Economic Viability High Upfront Costs Delayed ROI, below $100/kWh but still barrier Medium Material and manufacturing expenses Energy arbitrage, lifecycle optimization via RL 10-20 Declining costs but need for long-term operation [4], [19] Safety Hazards Internal Resistance Increase Heat generation, efficiency loss Medium High-frequency current ripples, variable loads AI-driven thermal management, ripple minimization 15-20 EV ripple effects extend to grids [3], [7] Integration Issues Compatibility with Infrastructure Increased complexity, highvoltage converter needs Medium Legacy grid systems, variable renewable inputs Hybrid AI-physical models, smart grid coordination 18-25 Microgrid buffering challenges [2], [10] Operational Efficiency Intermittency Buffering Failures Grid instability, curtailment (up to 25%) High Mismatched capacity to renewables AI forecasting and dispatch optimization 20-30 Wind/solar fluctuations [17], [18] Environmental Concerns Battery Disposal and Production Impact Higher carbon footprint if not optimized LowMedium Resource extraction, end-of-life management Lifespan extension via AI, recycling integration 10-18 Reduced replacements through better management [51], [52] Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 288 Cell-to-cell inconsistencies, arising from manufacturing tolerances or uneven aging, create imbalances in voltage and capacity within battery packs, reducing overall efficiency and lifespan. These imbalances are particularly problematic in large BESS, where series-parallel configurations amplify the impact of weak cells. Liu et al. [16] developed a model to quantify the impact of cell inconsistencies, finding that a 5% variation in initial capacity can lead to a 15% reduction in pack-level performance over 1000 cycles, underscoring the need for precise balancing algorithms in grid applications. Additionally, grid-scale LIB systems face integration challenges, such as compatibility with existing infrastructure and the need for high-voltage converters, which increase costs and complexity. Economic viability is another hurdle, as the high upfront cost of LIB systems, despite recent declines, requires long-term operation to achieve return on investment. Degradation and safety concerns further complicate cost-benefit analyses, particularly for utilities aiming to meet stringent reliability standards. These challenges highlight the necessity for advanced control systems, such as those leveraging AI, to mitigate degradation, enhance safety, and optimize performance in grid-scale deployments. Table 2 summarizes the primary challenges in deploying LIBs at grid scale, including their impacts, causes, and AI-enabled mitigation strategies, to highlight pathways for enhanced reliability. 2.3. Integration with Renewable Energy Systems LIBs play a pivotal role in integrating renewable energy sources, such as solar photovoltaic and wind, into power grids by storing surplus energy and releasing it during periods of low generation. This buffering capability addresses the intermittency of renewables, ensuring a stable power supply and reducing reliance on fossil fuel-based backup systems. In microgrids, LIBs enable islanded operation, supplying power to remote or critical loads during grid outages. A study by Hannan et al. [17] demonstrated that LIB-based BESS in hybrid renewable systems can reduce energy curtailment by up to 25%, enhancing grid reliability and supporting decarbonization goals. In grid-scale applications, LIBs provide ancillary services like frequency regulation and voltage support, leveraging their fast response times to stabilize grid parameters. For instance, during sudden drops in wind generation, LIBs can discharge stored energy within milliseconds to maintain frequency within the standard 50/60 Hz range. Diouf and Pode [18] reviewed the potential of LIBs in renewable energy storage, noting their ability to handle high-frequency power fluctuations, which is critical for grids with over 30% renewable penetration. This capability is particularly valuable in regions transitioning to net-zero energy systems, where renewables dominate the energy mix. The integration of LIBs with renewables also facilitates energy arbitrage, where batteries store low-cost energy during off-peak periods and discharge it during high-demand, high-price periods. This practice not only improves grid efficiency but also generates economic benefits for operators. According to Uddin et al. [19], LIBs in vehicle-to-grid (V2G) systems can achieve cost savings of up to $200 per vehicle annually through arbitrage, a model extensible to stationary BESS in renewable-heavy grids. However, effective integration requires precise control to align battery operations with grid demands, highlighting the role of AI in optimizing dispatch schedules. Despite their advantages, integrating LIBs with renewables faces challenges, such as matching battery capacity to variable generation profiles and managing degradation from irregular cycling patterns. These issues necessitate dynamic control strategies to balance energy input/output while preserving battery health. The synergy between LIBs and renewables, when optimized, can significantly enhance grid resilience and sustainability, paving the way for AIdriven solutions to address operational complexities. 3. AI Techniques for Battery Management Systems (BMS) The integration of artificial intelligence (AI) into battery management systems (BMS) has revolutionized the control and optimization of lithium-ion battery (LIB) performance, particularly in grid-scale applications. AI techniques, encompassing machine learning (ML), deep learning (DL), and reinforcement learning (RL), enable real-time analysis of complex battery data, such as voltage, current, and temperature, to enhance state estimation, thermal management, and operational efficiency. By leveraging predictive and adaptive algorithms, AI addresses the limitations of traditional BMS, such as rule-based controllers, offering dynamic solutions for grid integration and power supply optimization. This section explores key AI methodologies applied to BMS, focusing on state estimation, thermal management, and their impact on LIB reliability. 3.1. Overview of AI in BMS AI techniques have transformed BMS by enabling data-driven decision-making that adapts to the nonlinear and dynamic behavior of LIBs. Machine learning, including supervised and unsupervised algorithms, processes historical and realtime data to predict battery states and optimize control strategies. Deep learning, with its multi-layered neural Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 289 networks, excels in modeling complex electrochemical interactions, while reinforcement learning optimizes long-term performance through trial-and-error learning. These approaches collectively enhance BMS capabilities, from fault detection to energy management, making them critical for grid-scale applications. According to Chemali et al. [20], AIdriven BMS can achieve up to 98% accuracy in state predictions, significantly outperforming conventional methods like equivalent circuit models, which struggle with nonlinear dynamics. The application of ML in BMS includes algorithms like support vector machines (SVM), random forests, and neural networks, which analyze sensor data to predict parameters such as state of charge (SOC) and state of health (SOH). These algorithms are trained on datasets comprising thousands of charge-discharge cycles, enabling them to identify patterns that traditional models overlook. For instance, ML models can account for temperature and aging effects, reducing estimation errors to below 2% in real-world conditions. A comprehensive study by Hu et al. [21] reviewed ML applications in battery management, highlighting how supervised learning improves SOC estimation accuracy by 20– 30% compared to Kalman filter-based methods, particularly in dynamic grid scenarios. Deep learning, particularly recurrent neural networks (RNNs) and convolutional neural networks (CNNs), excels in processing time-series and multidimensional data, such as thermal profiles or impedance spectra. These models capture temporal dependencies and spatial variations in battery behavior, enabling precise fault diagnosis and predictive maintenance. RL, on the other hand, optimizes control policies for charge-discharge cycles by maximizing a reward function, such as energy efficiency or battery lifespan. Findings from Li et al. [22] demonstrate that RL-based BMS can extend battery life by 15% through adaptive charging strategies, a critical advantage for grid applications where longevity reduces operational costs. For a structured comparison, Table 3 outlines various AI techniques used in BMS, including their applications, inputs, benefits, and limitations, synthesizing insights from key studies. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 296 Table 5 Case Studies of AI-Enhanced Grid Integration with LIB Systems Case Study Location Scale (MW/MWh) AI Technique Used Key Objectives Achievements Challenges Faced Economic/Environmental Impact References Hornsdale Power Reserve South Australia 150/193.5 ML for Forecasting Frequency Control, Arbitrage 20% Cost Reduction, 100ms Response Data Latency Reduced Emissions by Displacing Fossil Fuels [38] Alameda County Microgrid California, USA 2/NA RL for Demand Management Peak Shaving, Efficiency 25% Demand Charge Reduction, 15% Efficiency Gain Integration with Solar Cost Savings, Lower Curtailment [39] Denmark Wind Farm Pilot Denmark 10/NA DL for Prediction Energy Dispatch Optimization 22% Curtailment Reduction, 90% Prediction Accuracy Variable Wind Patterns Enhanced Renewable Utilization [40] South Korea Hybrid Grid South Korea 5/10 Hybrid MLRL Voltage/Frequency Regulation 18% Voltage Deviation Reduction High EV Loads 15% Emission Reduction [37] European Energy Market BESS Europe Variable ML for Arbitrage Cost Minimization 20% Revenue Increase Market Volatility Decarbonization Support [49] California Solar Microgrid California, USA 1-5/Variable CNN for Coordination Power Loss Reduction 17% Loss Reduction Distributed Nodes Improved Sustainability [46] Australian V2G System Australia NA (EV Fleet) NN for V2G Grid Balancing $200/Vehicle Annual Savings Battery Degradation Reduced Peak Emissions [3], [19] Chinese Renewable Heavy Grid China 50/100 RL for Response Demand Response 30% Peak Stress Reduction Communication Delays 18% Carbon Reduction [36] Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 297 5. Power Supply Optimization Using AI The optimization of power supply in lithium-ion battery (LIB) systems is critical for maximizing efficiency, reducing costs, and supporting the integration of renewable energy into power grids. Artificial intelligence (AI) enhances power supply optimization by enabling precise control of charge-discharge cycles, improving energy dispatch strategies, and facilitating seamless coordination with smart grid technologies. Through machine learning (ML), deep learning (DL), and reinforcement learning (RL), AI algorithms analyze complex grid and battery data to achieve economic and environmental benefits. This section explores AI-driven approaches for energy management, smart grid integration, and the resulting economic and environmental impacts, highlighting their role in enhancing LIB performance for grid applications. 5.1. Energy Management and Efficiency Optimization AI-driven energy management systems (EMS) optimize the operation of LIB-based battery energy storage systems (BESS) by determining optimal charge-discharge schedules based on grid demands, energy prices, and battery health. ML algorithms, such as neural networks and decision trees, predict load profiles and renewable generation patterns, enabling BESS to store energy during low-cost periods and discharge during high-demand phases. This process, known as energy arbitrage, maximizes economic returns while minimizing battery degradation. According to Alam et al. [41], an ML-based EMS for BESS reduced energy costs by 18% in a microgrid by optimizing dispatch based on real-time price signals, demonstrating significant efficiency gains. Reinforcement learning is particularly effective for energy management, as it learns optimal control policies through interaction with dynamic grid environments. RL algorithms balance competing objectives, such as maximizing battery lifespan and meeting grid demands, by adjusting charge rates and discharge depths. In a comprehensive study, Hosseini et al. [42] implemented an RL-based EMS for LIBs in a renewable-heavy grid, achieving a 15% improvement in energy efficiency compared to rule-based controllers, primarily by reducing unnecessary cycling. Similarly, Cao et al. [43] demonstrated that RL could extend battery life by 12% through adaptive charging strategies that minimize capacity fade, critical for long-term grid reliability. AI also enhances efficiency by optimizing power conversion and thermal management within BESS. DL models predict heat generation during high-rate operations, enabling dynamic cooling adjustments that reduce energy losses. A study by Abbasi et al. [44] showed that a DL-based thermal optimization system reduced cooling energy consumption by 20% in a grid-scale BESS, improving overall system efficiency. Additionally, ML-driven state estimation ensures precise control of state of charge (SOC) and state of health (SOH), preventing overcharge or deep discharge that could compromise efficiency. These advancements collectively enable LIBs to deliver reliable power while minimizing operational costs. Challenges in AI-driven energy management include the computational complexity of real-time optimization and the need for high-quality data to train models. Cloud computing and edge-based AI solutions are addressing these issues, enabling scalable and efficient EMS for grid-scale applications. These innovations position AI as a key enabler of energy efficiency in LIB systems. 5.2. Integration with Smart Grids Smart grids, characterized by bidirectional communication and distributed energy resources, rely on AI-driven LIB systems to enhance power flow control and grid flexibility. AI algorithms coordinate BESS with smart grid components, such as inverters, demand response systems, and renewable generators, to optimize energy distribution. ML models predict grid dynamics, such as load fluctuations and voltage variations, enabling BESS to respond proactively to maintain stability. According to Jamil et al. [45], an ML-based control system for smart grid BESS reduced voltage fluctuations by 22% in a distributed network, enhancing power quality and supporting renewable integration. Deep learning enhances smart grid integration by processing multidimensional data, such as grid topology and realtime sensor inputs, to optimize energy dispatch. For instance, convolutional neural networks (CNNs) analyze spatialtemporal grid data to predict congestion points, allowing BESS to alleviate stress through targeted discharge. In a novel study, Srikanth et al. [46] applied a CNN-based approach to coordinate BESS in a smart grid, achieving a 17% reduction in power losses by optimizing energy flows across distributed nodes. Similarly, RL enables dynamic demand response, adjusting BESS operations to align with grid signals like price incentives or emergency load shedding. Nguyen [47] demonstrated that an RL-based BESS control system improved demand response efficiency by 25%, critical for smart grids with high EV penetration. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 298 AI also facilitates vehicle-to-grid (V2G) integration, where LIBs in electric vehicles contribute to smart grid stability by supplying power during peak demand. ML models predict EV charging patterns and grid needs, enabling coordinated V2G operations. A study by Noel et al. [48] found that AI-driven V2G systems increased grid reliability by 15% by optimizing bidirectional power flows, offering economic benefits for EV owners and grid operators. These capabilities highlight AI’s role in creating flexible, responsive smart grid ecosystems. The primary challenge in smart grid integration is ensuring interoperability between AI-driven BESS and diverse grid infrastructures. Standardized communication protocols and hybrid AI-physical models are emerging solutions, enabling seamless coordination across heterogeneous systems. These advancements ensure that LIBs maximize their contribution to smart grid performance. 5.3. Economic and Environmental Benefits AI-driven optimization of LIB systems yields significant economic benefits by reducing operational costs and improving energy market participation. Through energy arbitrage, BESS store low-cost energy during off-peak periods and sell it during high-price periods, generating revenue for operators. ML-based forecasting models enhance arbitrage profitability by predicting price trends with high accuracy. According to Sterner and Stadler [49], an AI-optimized BESS achieved a 20% increase in arbitrage revenue in a European energy market, demonstrating the economic potential of AI-driven systems. Additionally, predictive maintenance reduces downtime and replacement costs, further improving economic viability. A study by Talluri et al. [50] showed that ML-based maintenance strategies lowered BESS lifecycle costs by 12% through early fault detection. Environmentally, AI-driven LIB systems support decarbonization by maximizing renewable energy utilization and reducing reliance on fossil fuel-based peaking plants. By optimizing charge-discharge cycles, AI minimizes energy curtailment from renewables, ensuring efficient use of clean energy. In a comprehensive analysis, Arbabzadeh et al. [51] found that AI-optimized BESS reduced carbon emissions by 18% in a grid with 50% renewable penetration, primarily by displacing gas-fired generation. Furthermore, AI enhances battery lifespan, reducing the need for frequent replacements and minimizing the environmental impact of battery production and disposal. Madani et al. [52] reported that AI-driven charging strategies lowered the carbon footprint of BESS by 10% by extending battery life. AI also enables demand-side management, encouraging energy-efficient consumer behavior through dynamic pricing and load-shifting incentives. RL-based systems adjust BESS operations to align with low-carbon grid conditions, further reducing emissions. According to Kumari and Tanwar [53], an RL-driven demand response system reduced peak load emissions by 15% in a smart grid, supporting sustainability goals. These benefits position AI-driven LIB systems as key enablers of low-carbon energy transitions. Challenges include the high initial costs of AI implementation and the need for robust data to ensure accurate optimization. Continued advancements in cost-effective AI frameworks and open-access datasets will further amplify the economic and environmental benefits of AI-driven LIB systems in grid applications. 6. Challenges and Future Directions The application of artificial intelligence (AI) in lithium-ion battery (LIB) systems has significantly advanced grid integration and power supply optimization, yet several challenges persist in fully realizing its potential [54,56]. Issues such as data dependency, computational complexity, and scalability limit the widespread adoption of AI-driven battery management systems (BMS) in diverse grid environments. Addressing these challenges requires innovative approaches and interdisciplinary collaboration [57]. This section examines current limitations, emerging trends, and research gaps in AI-driven LIB control, providing a roadmap for future developments to enhance grid reliability and sustainability. 6.1. Current Limitations of AI in LIB Control AI-driven LIB control systems rely heavily on high-quality, large-scale datasets to train models for state estimation, thermal management, and energy optimization. Data scarcity or inconsistencies, particularly in real-world grid scenarios, can degrade model performance, leading to inaccurate predictions of state of charge (SOC) or state of health (SOH) [58-60]. For instance, variations in battery chemistry or operating conditions across grid-scale battery energy storage systems (BESS) complicate model generalization. According to Vidal et al. [61], limited access to comprehensive datasets reduces the accuracy of ML-based SOC estimation by up to 20% in heterogeneous battery packs, highlighting the need for robust data collection frameworks. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 299 Computational complexity poses another significant challenge, as AI models like deep neural networks (DNNs) and reinforcement learning (RL) require substantial processing power for real-time applications. In grid-scale BESS, where millisecond-level responses are critical for frequency regulation, the latency introduced by complex models can hinder performance. A study by Hu et al. [62] noted that DNN-based thermal management systems increased computation time by 30% compared to simpler models, necessitating trade-offs between accuracy and speed. This issue is particularly pronounced in resource-constrained environments, such as edge devices in microgrids. Scalability remains a barrier, as AI models developed for small-scale LIB systems often fail to adapt to large-scale BESS with thousands of cells. Cell-to-cell variations and grid-specific dynamics require models to be retrained or fine-tuned, increasing implementation costs. Findings from Berecibar et al. [63] indicate that scaling ML-based SOH prediction to utility-scale BESS reduced accuracy by 15% due to unmodeled variations in cell aging, underscoring the need for adaptive algorithms. Additionally, cybersecurity risks, such as data tampering in AI-driven BMS, pose threats to grid reliability, requiring robust protective measures. The integration of AI with existing grid infrastructure also faces interoperability challenges, as legacy systems may not support advanced communication protocols needed for real-time AI control. These limitations highlight the need for standardized frameworks and efficient computational strategies to ensure AI-driven LIB systems are practical and reliable in diverse grid applications. 6.2. Emerging Trends and Innovations Emerging trends in AI-driven LIB control are addressing current limitations through innovations like explainable AI (XAI), edge computing, and hybrid AI-physical models. XAI enhances model transparency by providing interpretable insights into AI decisions, which is crucial for gaining trust in grid operations. For instance, XAI can clarify why a model predicts a specific SOC or fault condition, aiding engineers in validating control strategies. According to Oyucu et al. [64], XAI-based BMS improved operator confidence in SOC predictions by 25%, facilitating adoption in safety-critical grid applications. Edge computing is revolutionizing AI implementation by enabling real-time processing on local devices, reducing latency and dependency on cloud infrastructure. This is particularly beneficial for microgrids and remote BESS, where connectivity may be limited. In a novel study, Minh et al. [65] demonstrated that edge-based ML models for thermal management reduced response times by 40% compared to cloud-based systems, enhancing efficiency in dynamic grid scenarios. Similarly, federated learning, where models are trained across distributed devices without sharing raw data, addresses privacy and bandwidth concerns. Wang et al. [66] applied federated learning to BESS, improving SOH estimation accuracy by 18% while maintaining data security. Hybrid AI-physical models combine data-driven AI with electrochemical or equivalent circuit models to improve accuracy and reduce data dependency. These models leverage physical principles to constrain AI predictions, ensuring robustness across operating conditions. A study by Chen et al. [67] showed that a hybrid DNN-electrochemical model reduced SOC estimation errors by 22% in variable temperature conditions, offering a scalable solution for grid-scale BESS. Additionally, advancements in transfer learning enable models trained on one battery type to adapt to others, addressing scalability challenges. Shu et al. [68] reported a 15% improvement in RUL prediction accuracy using transfer learning across different LIB chemistries. Quantum computing and neuromorphic hardware are also emerging, promising to accelerate AI computations for realtime BMS applications. These innovations, while in early stages, could transform LIB control by enabling complex optimizations with minimal latency. Continued development in these areas will enhance the efficiency and adaptability of AI-driven LIB systems. 6.3. Research Gaps and Opportunities Despite significant progress, several research gaps remain in AI-driven LIB control for grid integration. One key gap is the lack of standardized datasets for training AI models across diverse LIB chemistries and grid conditions. This limits model generalizability and hinders cross-system comparisons. According to Attia et al. [69], the absence of open-access, large-scale battery datasets reduces the reproducibility of AI models by 30%, necessitating collaborative efforts to create standardized data repositories. Such initiatives could accelerate research and deployment of AI-driven BMS. Real-time adaptability remains a challenge, as most AI models are trained offline and struggle to adapt to sudden changes in grid dynamics, such as unexpected renewable generation drops. Online learning techniques, which update models in real-time, offer a promising solution but require further development to balance accuracy and computational Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 300 efficiency. A study by She et al. [70] found that online ML models improved SOC estimation adaptability by 20% in dynamic grid scenarios, but computational demands limited their scalability. Research into lightweight online algorithms could address this gap. The integration of AI with cross-system grid components, such as distributed energy resources and demand response systems, is underexplored. Developing AI frameworks that coordinate BESS with other grid assets could enhance overall system efficiency. For instance, Smart et al. [71] highlighted a 25% efficiency gain in a microgrid using AI to coordinate BESS with solar inverters, suggesting opportunities for broader system-level optimization. Similarly, the environmental impact of AI-driven BMS, including the carbon footprint of model training, requires further investigation to ensure sustainability [72]. Opportunities also exist in developing AI models that incorporate multi-objective optimization, balancing battery lifespan, grid stability, and economic returns. RL-based multi-objective frameworks are gaining traction, as demonstrated by Kang et al. [73], who achieved a 15% improvement in combined economic and environmental outcomes. Future research should focus on these interdisciplinary approaches, leveraging advancements in AI, materials science, and grid engineering to unlock the full potential of LIB systems. 7. Conclusion The integration of artificial intelligence (AI) into lithium-ion battery (LIB) systems has marked a transformative shift in addressing the challenges of grid integration and power supply optimization. This review has demonstrated that AIdriven control strategies, encompassing machine learning, deep learning, and reinforcement learning, significantly enhance the performance of battery management systems by improving state estimation, thermal management, and energy dispatch efficiency. These advancements enable LIBs to effectively support renewable energy integration, stabilize grid operations, and reduce operational costs, positioning them as critical components in the transition to sustainable energy ecosystems. The ability of AI to process complex, real-time data ensures that LIB systems operate reliably under dynamic grid conditions, addressing issues such as capacity fade, thermal runaway, and cell inconsistencies. The application of AI in grid-scale battery energy storage systems facilitates seamless coordination with smart grid technologies, enabling precise frequency regulation, voltage control, and demand response. Case studies, such as the Horns dale Power Reserve and microgrid implementations, illustrate the practical benefits of AI, including reduced energy curtailment and enhanced economic returns through energy arbitrage. Moreover, AI-driven optimization yields substantial environmental benefits by maximizing renewable energy utilization and minimizing carbon emissions, aligning with global decarbonization goals. These outcomes underscore the pivotal role of AI in unlocking the full potential of LIBs for modern power systems. Despite these advancements, challenges such as data dependency, computational complexity, and scalability persist, necessitating continued research into explainable AI, edge computing, and hybrid models to ensure robust and adaptable solutions. The identified research gaps, including the need for standardized datasets and real-time adaptability, present opportunities for innovation that could further enhance grid integration and system efficiency. As AI technologies evolve, their integration with LIB systems promises to drive the development of resilient, cost-effective, and sustainable power grids. This review advocates for sustained interdisciplinary efforts to address existing limitations and fully realize the transformative impact of AI-driven LIB control, paving the way for a future where clean energy is both reliable and accessible. Compliance with ethical standards Acknowledgments The authors wish to acknowledge the collaborative effort of all contributing scholars and colleagues who jointly authored and edited this review paper. This work was conducted entirely through the intellectual and academic contributions of the authoring team, without external funding or assistance from any individual, institution, or organization. Disclosure of conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 301 References [1] Stroe, D. I., Swierczynski, M., Stan, A. I., Knap, V., Teodorescu, R., and Andreasen, S. J. (2014, September). Diagnosis of lithium-ion batteries state-of-health based on electrochemical impedance spectroscopy technique. In 2014 IEEE Energy Conversion Congress and Exposition (ECCE) (pp. 4576-4582). IEEE. [2] Khalid, A., Stevenson, A., and Sarwat, A. I. (2021). Overview of technical specifications for grid-connected microgrid battery energy storage systems. IEEe Access, 9, 163554-163593. [3] Uddin, K., Jackson, T., Widanage, W. D., Chouchelamane, G., Jennings, P. A., and Marco, J. (2017). On the possibility of extending the lifetime of lithium-ion batteries through optimal V2G facilitated by an integrated vehicle and smart-grid system. Energy, 133, 710-722. [4] Hakam, Y., Gaga, A., Tabaa, M., and Elhadadi, B. (2024). Intelligent integration of vehicle-to-grid (V2G) and vehiclefor-grid (V4G) systems: leveraging artificial neural networks (ANNs) for smart grid. Energies, 17(13), 3095. [5] Lipu, M. H., Miah, M. S., Jamal, T., Rahman, T., Ansari, S., Rahman, M. S., ... and Shakib, M. N. (2023). Artificial intelligence approaches for advanced battery management system in electric vehicle applications: A statistical analysis towards future research opportunities. Vehicles, 6(1), 22-70. [6] Boujoudar, Y., Elmoussaoui, H., and Lamhamdi, T. (2019). Lithium-ion batteries modeling and state of charge estimation using artificial neural network. International Journal of Electrical and Computer Engineering, 9(5), 3415. [7] Wang, X., Li, M., Liu, Y., Sun, W., Song, X., and Zhang, J. (2017). Surrogate based multidisciplinary design optimization of lithium-ion battery thermal management system in electric vehicles. Structural and Multidisciplinary Optimization, 56(6), 1555-1570. [8] Wali, S. B., Hannan, M. A., Ker, P. J., Abd Rahman, M. S., Mansor, M., Muttaqi, K. M., ... and Begum, R. A. (2022). Gridconnected lithium-ion battery energy storage system: A bibliometric analysis for emerging future directions. Journal of Cleaner Production, 334, 130272. [9] Oyucu, S., Ersöz, B., Sağıroğlu, Ş., Aksöz, A., and Biçer, E. (2024). Optimizing Lithium-ion battery performance: Integrating machine learning and explainable AI for enhanced energy management. Sustainability, 16(11), 4755. [10] Gjelaj, M., Hashemi, S., Traeholt, C., and Andersen, P. B. (2018). Grid integration of DC fast‐charging stations for EVs by using modular li‐ion batteries. IET Generation, Transmission and Distribution, 12(20), 4368-4376. [11] Kim, H. J., Krishna, T. N. V., Zeb, K., Rajangam, V., Gopi, C. V. M., Sambasivam, S., ... and Obaidat, I. M. (2020). A comprehensive review of Li-ion battery materials and their recycling techniques. Electronics, 9(7), 1161. [12] Kalhoff, J., Eshetu, G. G., Bresser, D., and Passerini, S. (2015). Safer electrolytes for lithium‐ion batteries: state of the art and perspectives. ChemSusChem, 8(13), 2154-2175. [13] Aurbach, D., Zinigrad, E., Cohen, Y., and Teller, H. (2002). A short review of failure mechanisms of lithium metal and lithiated graphite anodes in liquid electrolyte solutions. Solid state ionics, 148(3-4), 405-416. [14] Luo, G., Zhang, Y., and Tang, A. (2023). Capacity degradation and aging mechanisms evolution of lithium-ion batteries under different operation conditions. Energies, 16(10), 4232. [15] Feng, X., Ouyang, M., Liu, X., Lu, L., Xia, Y., and He, X. (2018). Thermal runaway mechanism of lithium ion battery for electric vehicles: A review. Energy storage materials, 10, 246-267. [16] Liu, K., Li, K., Peng, Q., and Zhang, C. (2019). A brief review on key technologies in the battery management system of electric vehicles. Frontiers of mechanical engineering, 14(1), 47-64. [17] Hannan, M. A., Hoque, M. M., Mohamed, A., and Ayob, A. (2017). Review of energy storage systems for electric vehicle applications: Issues and challenges. Renewable and Sustainable Energy Reviews, 69, 771-789. [18] Diouf, B., and Pode, R. (2015). Potential of lithium-ion batteries in renewable energy. Renewable Energy, 76, 375380. [19] Uddin, K., Dubarry, M., and Glick, M. B. (2018). The viability of vehicle-to-grid operations from a battery technology and policy perspective. Energy policy, 113, 342-347. [20] Chemali, E., Kollmeyer, P. J., Preindl, M., and Emadi, A. (2018). State-of-charge estimation of Li-ion batteries using deep neural networks: A machine learning approach. Journal of Power Sources, 400, 242-255. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 302 [21] Hu, X., Zhang, K., Liu, K., Lin, X., Dey, S., and Onori, S. (2020). Advanced fault diagnosis for lithium-ion battery systems: A review of fault mechanisms, fault features, and diagnosis procedures. IEEE Industrial Electronics Magazine, 14(3), 65-91. [22] Li, W., Sengupta, N., Dechent, P., Howey, D., Annaswamy, A., and Sauer, D. U. (2021). Online capacity estimation of lithium-ion batteries with deep long short-term memory networks. Journal of power sources, 482, 228863. [23] Hannan, M. A., Lipu, M. S. H., Hussain, A., Saad, M. H., and Ayob, A. (2018). Neural network approach for estimating state of charge of lithium-ion battery using backtracking search algorithm. Ieee Access, 6, 10069-10079. [24] Severson, K. A., Attia, P. M., Jin, N., Perkins, N., Jiang, B., Yang, Z., ... and Braatz, R. D. (2019). Data-driven prediction of battery cycle life before capacity degradation. Nature Energy, 4(5), 383-391. [25] Wu, H., Niu, M., Shao, Y., Wang, M., Li, M., Liu, X., and Li, Z. (2023). Investigations of li-ion battery thermal management systems based on heat pipes: A review. ACS omega, 9(1), 97-116. [26] Nuhic, A., Terzimehic, T., Soczka-Guth, T., Buchholz, M., and Dietmayer, K. (2013). Health diagnosis and remaining useful life prognostics of lithium-ion batteries using data-driven methods. Journal of power sources, 239, 680688. [27] Zhang, Y., Xiong, R., He, H., and Pecht, M. G. (2018). Long short-term memory recurrent neural network for remaining useful life prediction of lithium-ion batteries. IEEE Transactions on Vehicular Technology, 67(7), 5695-5705. [28] Ma, G., Zhang, Y., Cheng, C., Zhou, B., Hu, P., and Yuan, Y. (2019). Remaining useful life prediction of lithium-ion batteries based on false nearest neighbors and a hybrid neural network. Applied Energy, 253, 113626. [29] Abbasi, M. H., Arjmandzadeh, Z., Zhang, J., Xu, B., and Krovi, V. (2024). Deep reinforcement learning based fast charging and thermal management optimization of an electric vehicle battery pack. Journal of Energy Storage, 95, 112466. [30] Jaguemont, J., Boulon, L., and Dubé, Y. (2016). A comprehensive review of lithium-ion batteries used in hybrid and electric vehicles at cold temperatures. Applied Energy, 164, 99-114. [31] Samanta, A., Chowdhuri, S., and Williamson, S. S. (2021). Machine learning-based data-driven fault detection/diagnosis of lithium-ion battery: A critical review. Electronics, 10(11), 1309. [32] Areola, R. I., Adebiyi, A. A., and Moloi, K. (2025). Integrated Energy Storage Systems for Enhanced Grid Efficiency: A Comprehensive Review of Technologies and Applications. Energies (19961073), 18(7). [33] Duan, J., Yi, Z., Shi, D., Lin, C., Lu, X., and Wang, Z. (2019). Reinforcement-learning-based optimal control of hybrid energy storage systems in hybrid AC–DC microgrids. IEEE Transactions on Industrial Informatics, 15(9), 53555364. [34] Li, Y., Liu, K., Foley, A. M., Zülke, A., Berecibar, M., Nanini-Maury, E., ... and Hoster, H. E. (2019). Data-driven health estimation and lifetime prediction of lithium-ion batteries: A review. Renewable and sustainable energy reviews, 113, 109254. [35] Rosewater, D., Ferreira, S., Schoenwald, D., Hawkins, J., and Santoso, S. (2018). Battery energy storage state-ofcharge forecasting: Models, optimization, and accuracy. IEEE Transactions on Smart Grid, 10(3), 2453-2462. [36] Cao, J., Harrold, D., Fan, Z., Morstyn, T., Healey, D., and Li, K. (2020). Deep reinforcement learning-based energy storage arbitrage with accurate lithium-ion battery degradation model. IEEE Transactions on Smart Grid, 11(5), 4513-4521. [37] Zhang, J., Li, Y., Wu, Z., Rong, C., Wang, T., Zhang, Z., and Zhou, S. (2021). Deep-reinforcement-learning-based twotimescale voltage control for distribution systems. Energies, 14(12), 3540. [38] Kebede, A. A., Kalogiannis, T., Van Mierlo, J., and Berecibar, M. (2022). A comprehensive review of stationary energy storage devices for large scale renewable energy sources grid integration. Renewable and sustainable energy reviews, 159, 112213. [39] Hosseini, E., Horrillo-Quintero, P., Carrasco-Gonzalez, D., Garcia-Trivino, P., Sarrias-Mena, R., Garcia-Vazquez, C. A., and Fernandez-Ramirez, L. M. (2025). Reinforcement learning-based energy management system for lithiumion battery storage in multilevel microgrid. Journal of Energy Storage, 109, 115114. [40] Gjelaj, M., Hashemi, S., Andersen, P. B., and Traeholt, C. (2020). Optimal infrastructure planning for EV fast‐ charging stations based on prediction of user behaviour. IET Electrical Systems in Transportation, 10(1), 1-12. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 303 [41] Alam, M. M., Rahman, M. H., Ahmed, M. F., Chowdhury, M. Z., and Jang, Y. M. (2022). Deep learning based optimal energy management for photovoltaic and battery energy storage integrated home micro-grid system. Scientific reports, 12(1), 15133. [42] Hosseini, E., Horrillo-Quintero, P., Carrasco-Gonzalez, D., Garcia-Trivino, P., Sarrias-Mena, R., Garcia-Vazquez, C. A., and Fernandez-Ramirez, L. M. (2025). Reinforcement learning-based energy management system for lithiumion battery storage in multilevel microgrid. Journal of Energy Storage, 109, 115114. [43] Cao, J., Harrold, D., Fan, Z., Morstyn, T., Healey, D., and Li, K. (2020). Deep reinforcement learning-based energy storage arbitrage with accurate lithium-ion battery degradation model. IEEE Transactions on Smart Grid, 11(5), 4513-4521. [44] Abbasi, M. H., Arjmandzadeh, Z., Zhang, J., Xu, B., and Krovi, V. (2024). Deep reinforcement learning based fast charging and thermal management optimization of an electric vehicle battery pack. Journal of Energy Storage, 95, 112466. [45] Jamil, F., Iqbal, N., Ahmad, S., and Kim, D. (2021). Peer-to-peer energy trading mechanism based on blockchain and machine learning for sustainable electrical power supply in smart grid. Ieee Access, 9, 39193-39217. [46] Srikanth, D., Sukumar, G. D., and Sobhan, P. V. (2025). A convolutional neural network based energy management system for photovoltaic/battery systems in microgrid using enhanced coati optimization approach. Journal of Energy Storage, 119, 116252. [47] Nguyen, D. H. (2024). The Role of Reinforcement Learning Control for Optimizing Building Energy Management Systems. [48] Noel, L., Brodie, J. F., Kempton, W., Archer, C. L., and Budischak, C. (2017). Cost minimization of generation, storage, and new loads, comparing costs with and without externalities. Applied Energy, 189, 110-121. [49] Sterner, M., and Stadler, I. (Eds.). (2019). Handbook of energy storage: Demand, technologies, integration. Springer. [50] Talluri, G., Lozito, G. M., Grasso, F., Iturrino Garcia, C., and Luchetta, A. (2021). Optimal battery energy storage system scheduling within renewable energy communities. Energies, 14(24), 8480. [51] Arbabzadeh, M., Sioshansi, R., Johnson, J. X., and Keoleian, G. A. (2019). The role of energy storage in deep decarbonization of electricity production. Nature communications, 10(1), 3413. [52] Madani, S. S., Shabeer, Y., Fowler, M., Panchal, S., Chaoui, H., Mekhilef, S., ... and See, K. (2025). Artificial Intelligence and Digital Twin Technologies for Intelligent Lithium-Ion Battery Management Systems: A Comprehensive Review of State Estimation, Lifecycle Optimization, and Cloud-Edge Integration. Batteries, 11(8), 298. [53] Kumari, A., and Tanwar, S. (2021). A reinforcement-learning-based secure demand response scheme for smart grid system. IEEE Internet of Things Journal, 9(3), 2180-2191. [54] Zheng, Y., Zhao, J., Song, Y., Luo, F., Meng, K., Qiu, J., and Hill, D. J. (2017). Optimal operation of battery energy storage system considering distribution system uncertainty. IEEE Transactions on Sustainable Energy, 9(3), 1051-1060. [55] Lee, J., Kim, J. M., Yi, J., and Won, C. Y. (2021). Battery management system algorithm for energy storage systems considering battery efficiency. Electronics, 10(15), 1859. [56] Giglio, E., Luzzani, G., Terranova, V., Trivigno, G., Niccolai, A., and Grimaccia, F. (2023). An efficient artificial intelligence energy management system for urban building integrating photovoltaic and storage. IEEE Access, 11, 18673-18688. [57] Yang, K., Li, C., Jing, X., Zhu, Z., Wang, Y., Ma, H., and Zhang, Y. (2022). Energy dispatch optimization of islanded multi-microgrids based on symbiotic organisms search and improved multi-agent consensus algorithm. Energy, 239, 122105. [58] Van Hentenryck, P. (2021). Machine learning for optimal power flows. Tutorials in Operations Research: Emerging Optimization Methods and Modeling Techniques with Applications, 62-82. [59] Egbuna, I. K., Salihu, F. B., Okara, C. C., Olayiwola, D. E., Smart, E. E., Anifowose, O., and Mbamalu, P. O. (2025). Advances in AI-powered energy management systems for renewable-integrated smart grids. World Journal of Advanced Engineering Technology and Sciences, 15(2), 2300-2325. Global Journal of Engineering and Technology Advances, 2025, 24(03), 281-304 304 [60] Carter, E. (2023). AI-Optimized Energy Storage Systems for HighEfficiency Renewable Energy Integration. International Journal of AI, BigData, Computational and Management Studies, 4(1), 1-10. [61] Vidal, C., Malysz, P., Kollmeyer, P., and Emadi, A. (2020). Machine learning applied to electrified vehicle battery state of charge and state of health estimation: State-of-the-art. Ieee Access, 8, 52796-52814. [62] Hu, X., Li, S. E., and Yang, Y. (2015). Advanced machine learning approach for lithium-ion battery state estimation in electric vehicles. IEEE Transactions on Transportation electrification, 2(2), 140-149. [63] Berecibar, M., Gandiaga, I., Villarreal, I., Omar, N., Van Mierlo, J., and Van den Bossche, P. (2016). Critical review of state of health estimation methods of Li-ion batteries for real applications. Renewable and Sustainable Energy Reviews, 56, 572-587. [64] Oyucu, S., Ersöz, B., Sağıroğlu, Ş., Aksöz, A., and Biçer, E. (2024). Optimizing Lithium-ion battery performance: Integrating machine learning and explainable AI for enhanced energy management. Sustainability, 16(11), 4755. [65] Minh, Q. N., Nguyen, V. H., Quy, V. K., Ngoc, L. A., Chehri, A., and Jeon, G. (2022). Edge computing for IoT-enabled smart grid: The future of energy. Energies, 15(17), 6140. [66] Wang, X., Liang, Y., Chen, Y., and Dobre, O. A. (2024, April). Robust Federated Learning for Energy Storage Systems. In 2024 IEEE Wireless Communications and Networking Conference (WCNC) (pp. 01-06). IEEE. [67] Chen, G., Jiang, S., Xie, M., and Yang, F. (2022, October). A hybrid DNN-KF model for real-time SOC estimation of lithium-ion batteries under different ambient temperatures. In 2022 Global Reliability and Prognostics and Health Management (PHM-Yantai) (pp. 1-5). IEEE.. [68] Shu, H., Li, W., Wang, G., Han, Y., Li, J., and Tang, Y. (2024). Online collaborative estimation technology for SOC and SOH of frequency regulation of a lead-carbon battery in a power system with a high proportion of renewable energy. Protection and Control of Modern Power Systems, 9(1), 52-64. [69] Attia, P. M., Grover, A., Jin, N., Severson, K. A., Markov, T. M., Liao, Y. H., ... and Chueh, W. C. (2020). Closed-loop optimization of fast-charging protocols for batteries with machine learning. Nature, 578(7795), 397-402. [70] She, C., Li, Y., Zou, C., Wik, T., Wang, Z., and Sun, F. (2021). Offline and online blended machine learning for lithiumion battery health state estimation. IEEE Transactions on Transportation Electrification, 8(2), 1604-1618. [71] Smart, E. E., Otaru, K., Raji, K. O., Asere, J. B., Anifowose, O., Fajobi, M. O., and Ojo, O. T. (2025). Sustaining the power grid: Breakthroughs and challenges in long-duration energy storage. [72] Khan, N., Shahid, Z., Alam, M. M., Bakar Sajak, A. A., Mazliham, M. S., Khan, T. A., and Ali Rizvi, S. S. (2022). Energy management systems using smart grids: an exhaustive parametric comprehensive analysis of existing trends, significance, opportunities, and challenges. International Transactions on Electrical Energy Systems, 2022(1), 3358795. [73] Kang, H., Jung, S., Kim, H., Hong, J., Jeoung, J., and Hong, T. (2023). Multi-objective sizing and real-time scheduling of battery energy storage in energy-sharing community based on reinforcement learning. Renewable and sustainable energy reviews, 185, 113655.