A Hybrid Machine Learning Framework for Accurate SOC and SOH Estimation in Lithium-Ion Batteries Using Matlab

Author : Waqas Javaid

Abstract

This paper presents a machine learning based smart battery management system for accurate state-of-charge (SOC) and state-of-health (SOH) estimation of lithium-ion batteries operating under dynamic load conditions [1]. A physics-informed second-order RC equivalent circuit model is employed to simulate realistic battery voltage and degradation behavior. Dynamic current profiles resembling real driving conditions are used to generate training and testing datasets. An LSTM neural network is developed to estimate SOC by learning temporal dependencies between current, terminal voltage, and battery state [2]. SOH is estimated through a capacity-fade model fused with SOC variation and charge throughput information. The proposed framework effectively captures nonlinear battery dynamics and long-term degradation trends. Simulation results demonstrate close agreement between estimated and true SOC and SOH values. The SOC estimation error remains low across varying load conditions, indicating strong generalization capability [3]. Performance evaluation using RMSE confirms the accuracy and robustness of the proposed approach. The results validate the suitability of machine learning–driven BMS architectures for intelligent energy storage management.

1. Introduction

The rapid growth of electric vehicles, renewable energy storage systems, and portable electronics has significantly increased the demand for efficient and reliable battery management systems.

Lithium-ion batteries are widely adopted due to their high energy density, long cycle life, and favorable performance characteristics. However, their safe and efficient operation strongly depends on accurate estimation of the state-of-charge and state-of-health, which are critical indicators of available energy and battery aging. Conventional SOC and SOH estimation techniques, such as Coulomb counting and model-based observers, often suffer from accumulated errors, parameter sensitivity, and limited robustness under dynamic operating conditions [4]. Real-world battery usage is highly nonlinear and time-varying, making accurate state estimation a challenging task. In recent years, machine learning techniques have emerged as powerful tools for modeling complex nonlinear systems without requiring explicit physical equations. Among these techniques, long short-term memory networks are particularly suitable for battery applications due to their ability to capture temporal dependencies in sequential data. By leveraging voltage and current measurements, LSTM models can effectively learn hidden electrochemical behaviors [5]. Additionally, data-driven degradation modeling enables reliable SOH estimation by tracking capacity loss over time. Integrating machine learning with physics-based battery models enhances estimation accuracy and system reliability. This study proposes an intelligent battery management framework that combines LSTM-based SOC estimation with capacity-based SOH prediction under dynamic load conditions [6]. The proposed approach aims to improve robustness, accuracy, and adaptability of modern battery management systems for real-time applications.

1.1 Background and Motivation

The increasing deployment of electric vehicles, renewable energy storage systems, and portable electronic devices has intensified the need for advanced battery management systems. Lithium-ion batteries are preferred in these applications because of their high energy density, long cycle life, and relatively low self-discharge rate [7]. Despite these advantages, lithium-ion batteries are highly sensitive to operating conditions such as load variations, temperature, and aging effects. Accurate knowledge of the state-of-charge and state-of-health is essential to ensure safe operation, prevent overcharging or deep discharging, and maximize battery lifespan. Traditional estimation methods rely on simplified mathematical models or direct integration of current, which often leads to accumulated errors. These methods also require frequent recalibration and precise parameter identification [8]. Under dynamic load conditions, their performance degrades significantly. Therefore, there is a strong motivation to develop more robust and adaptive estimation techniques. Intelligent estimation strategies are essential for next-generation battery management systems.

1.2 Limitations of Conventional Approaches

Conventional SOC estimation techniques such as Coulomb counting, open-circuit voltage lookup, and Kalman filter–based observers are widely used in practical systems. However, Coulomb counting is highly sensitive to sensor noise and initial condition errors. Open-circuit voltage methods require long rest periods, which are impractical during real-time operation. Model-based observers depend on accurate battery models whose parameters change with aging and temperature [9]. As batteries degrade, these models lose accuracy and require re-identification. Furthermore, nonlinear electrochemical dynamics are difficult to capture using low-order equivalent circuit models alone. These limitations restrict the applicability of traditional approaches in real-world scenarios. As a result, estimation accuracy deteriorates under highly dynamic load profiles. This motivates the exploration of data-driven and adaptive methods.

1.3 Role of Machine Learning in Battery Management

Machine learning has emerged as a promising solution for handling nonlinear and time-varying systems. Unlike physics-based models, machine learning techniques can learn complex relationships directly from measured data.

Table 1: Battery Parameters

Neural networks, in particular, have shown strong performance in battery state estimation tasks. Long short-term memory networks are well suited for time-series data due to their internal memory structure [10]. They can capture long-term dependencies between voltage, current, and battery states that are difficult to model analytically. LSTM networks are also resilient to noise and measurement uncertainty. By training on historical operational data, these models can generalize well to unseen conditions. This makes machine learning an attractive tool for intelligent battery management.

1.4 SOC Estimation Using LSTM Networks

Accurate SOC estimation remains one of the most critical challenges in battery management. SOC is not directly measurable and must be inferred from available signals such as current and terminal voltage. LSTM-based SOC estimation leverages sequential input data to learn the underlying battery dynamics [11]. By considering both short-term fluctuations and long-term trends, LSTM networks outperform traditional feedforward neural networks. They reduce estimation drift under dynamic load conditions and varying operating states. Moreover, LSTM models can adapt to changing battery characteristics over time. This improves robustness in real-world applications such as electric vehicles and grid storage systems [12]. The use of deep learning enables higher estimation accuracy without complex parameter tuning.

1.5 SOH Estimation and Degradation Modeling

State-of-health estimation provides insight into battery aging and remaining useful life. SOH is commonly associated with capacity fade and internal resistance growth. Accurate SOH estimation allows predictive maintenance and prevents unexpected battery failures. In this work, SOH is estimated using a capacity-based degradation approach fused with SOC variations. By analyzing charge throughput over a moving window, battery capacity loss can be inferred. This method complements the LSTM-based SOC estimator by providing long-term health information [13]. The integration of SOC and SOH estimation results in a comprehensive smart battery management framework. Such an approach enhances reliability, safety, and lifetime optimization of lithium-ion battery systems.

2. Problem Statement

Accurate estimation of the state-of-charge and state-of-health of lithium-ion batteries remains a critical challenge for modern battery management systems operating under dynamic load conditions. SOC and SOH are internal states that cannot be directly measured and must be inferred from noisy voltage and current signals. Conventional estimation methods suffer from accumulated errors, sensitivity to model parameters, and reduced accuracy as batteries age. Rapid load fluctuations further degrade the performance of model-based and rule-based techniques. Battery degradation introduces nonlinear and time-varying behavior that is difficult to capture using fixed-parameter models. Inaccurate state estimation can lead to unsafe operation, reduced usable capacity, and shortened battery lifespan. Existing approaches often treat SOC and SOH estimation separately, limiting overall system reliability. There is a lack of integrated frameworks capable of simultaneously handling short-term dynamics and long-term degradation. Additionally, many methods require extensive calibration and are not easily adaptable to real-world conditions. Therefore, a robust, adaptive, and data-driven solution is required to ensure reliable SOC and SOH estimation for intelligent battery management systems.

3. Mathematical Approach

The mathematical framework of the proposed smart battery management system combines physics-based battery modeling with data-driven state estimation. A second-order RC equivalent circuit model is used to represent the electrical behavior of the lithium-ion battery, where the terminal voltage is expressed as the open-circuit voltage minus voltage drops across internal resistances and RC networks. The open-circuit voltage is modeled as a nonlinear function of the state-of-charge. SOC dynamics are governed by Coulomb counting, where SOC decreases proportionally to the applied current and sampling time normalized by the effective battery capacity. Battery capacity is modeled as a function of state-of-health, allowing SOC dynamics to reflect degradation effects. SOH evolution is represented using a gradual capacity fade model proportional to charge throughput [14]. The RC network dynamics are described by first-order differential equations capturing transient polarization effects. These continuous-time equations are discretized using a fixed sampling interval. The resulting voltage and current data form the observation vector for the learning model. An LSTM network is trained to approximate the nonlinear mapping between sequential inputs and SOC output. The LSTM learns temporal dependencies through gated memory cells and backpropagation through time. Mean squared error is used as the loss function during training. SOH is estimated by relating the change in estimated SOC over a sliding window to the integrated current, yielding an effective capacity estimate. The estimated capacity is normalized with respect to nominal capacity to obtain SOH. This hybrid mathematical approach ensures both short-term accuracy and long-term degradation awareness.

The state-of-charge evolution is mathematically expressed as:

Where, (I(k) ) is the battery current, ( T_s ) is the sampling time, and (Q_nom) is the nominal capacity. The terminal voltage is given by:

With RC dynamics:

The capacity degradation is modeled as:

Where, (Q_est) is obtained from charge throughput. The LSTM learns the nonlinear mapping by minimizing the mean squared error between true and estimated SOC.

The state-of-charge equation describes how the battery SOC decreases over time as a function of the applied current, sampling interval, and effective battery capacity adjusted by state-of-health. This formulation ensures that capacity degradation directly influences the rate of SOC change, making the model sensitive to aging effects. The terminal voltage equation represents the measured battery voltage as the open-circuit voltage reduced by ohmic losses and transient polarization effects. The RC voltage equations capture the dynamic behavior of the battery during current transients by modeling diffusion and charge transfer phenomena. Finally, the learned mapping in the LSTM network estimates SOC by identifying nonlinear temporal relationships between current, voltage, and battery state from historical data.

You can download the Project files here: Download files now. (You must be logged in).

4. Methodology

The proposed methodology follows a hybrid modeling and data-driven framework for intelligent battery state estimation. First, a second-order RC equivalent circuit model is employed to simulate realistic lithium-ion battery voltage dynamics under varying load conditions. A dynamic current profile is generated to emulate real-world operating scenarios and to excite both short-term and long-term battery behaviors.

Table 2: SOC & SOH Comparison

True SOC and SOH trajectories are computed using Coulomb counting and a capacity fade model, providing ground truth data for supervised learning [15]. Voltage and current measurements are preprocessed and normalized to improve learning stability. The dataset is then divided into training and testing subsets to ensure unbiased performance evaluation. An LSTM neural network is designed to estimate SOC by learning temporal dependencies from sequential input data. The network is trained using the Adam optimizer with mean squared error as the loss function. After training, the LSTM model is used to estimate SOC under unseen test conditions. SOC estimation errors are computed to assess accuracy and robustness [16]. For SOH estimation, a capacity-based approach is applied using estimated SOC variations and integrated current over a sliding time window. This enables the extraction of effective battery capacity and corresponding health state. The SOC and SOH estimators are fused within a unified smart battery management system. Finally, multiple performance metrics and graphical analyses are used to validate the effectiveness of the proposed methodology under dynamic load conditions [17].

5. Design Matlab Simulation and Analysis

The simulation begins by defining the battery and system parameters, including sampling time, simulation duration, and nominal battery characteristics such as capacity, voltage, and internal resistances.

Table 3: Simulation Parameters

A dynamic load profile is generated to mimic real-world operating conditions, combining sinusoidal variations with random noise, and current values are constrained to realistic operating limits [18]. The true battery states, including SOC and SOH, are simulated using a second-order RC equivalent circuit model, where SOC is updated using Coulomb counting adjusted by the capacity fade model, and SOH gradually decreases based on cumulative current usage. Transient RC voltages are computed to capture polarization effects, while the open-circuit voltage is modeled as a nonlinear function of SOC. Terminal voltage is calculated by combining open-circuit voltage, ohmic drop, and RC contributions, providing realistic measurement data for machine learning. The voltage and current data are then normalized and divided into training and testing datasets to ensure unbiased LSTM network evaluation. The LSTM network is designed with sequence input, a hidden layer of 50 units, and a regression output layer to estimate SOC from sequential voltage and current data [19]. Training is performed using the Adam optimizer for 60 epochs with a batch size of 64, and training progress is visualized. After training, the LSTM predicts SOC on the test dataset, and estimation errors are computed. SOH estimation is performed using a sliding window approach, relating changes in SOC to integrated current to infer effective battery capacity. Estimated capacity is normalized to obtain SOH and compared with true SOH for validation. Seven output figures are generated, illustrating the current profile, terminal voltage, SOC estimation, SOC error, SOH estimation, capacity degradation, and SOC versus voltage relationship. RMSE is calculated to quantify SOC estimation accuracy. Overall, the simulation demonstrates the ability of the hybrid LSTM and degradation model to capture both short-term battery dynamics and long-term aging trends under realistic load conditions.

This figure shows the applied current profile used to simulate realistic operating conditions. The current combines low-frequency and high-frequency sinusoidal components to represent typical driving or load cycles. Random noise is added to emulate measurement uncertainty and real-world variations. Negative currents are set to zero, reflecting the charging or discharge constraints of the battery. Peak currents are limited to a maximum safe value to prevent unrealistic operating conditions. The dynamic nature of the current excites both short-term transient and long-term battery dynamics. Observing this profile helps to understand how the battery responds under variable load. It provides the input for SOC and SOH estimation. The figure highlights the fluctuating demands that a smart battery management system must handle. Overall, it sets the foundation for evaluating the performance of the ML-based estimation approach.

This figure depicts the terminal voltage of the battery calculated using the second-order RC model. It reflects the combined effects of open-circuit voltage, ohmic resistance, and RC transient voltages. The voltage fluctuates according to changes in the applied current, showing dips during high current demand and rises during low current intervals. It captures the nonlinear response of the battery, which is critical for accurate SOC estimation. Voltage measurements are affected by the battery’s internal resistance and polarization effects. The figure provides insight into the electrical behavior of the battery under dynamic operation. It serves as one of the main inputs for the LSTM network. Observing voltage trends helps identify periods of high stress or capacity utilization. This data is crucial for training the machine learning model.

You can download the Project files here: Download files now. (You must be logged in).

This figure shows the true SOC versus the SOC estimated by the trained LSTM network for the test dataset. The true SOC is calculated using Coulomb counting adjusted by capacity fade, while the LSTM prediction is based on sequential voltage and current inputs. The estimated SOC closely follows the true SOC, demonstrating the network’s ability to capture nonlinear battery dynamics. Small deviations highlight transient estimation errors during rapid current changes. The figure demonstrates the model’s generalization capability under unseen load profiles. It emphasizes the effectiveness of LSTM in handling temporal dependencies in sequential battery data. Accurate SOC estimation is critical for safe battery operation and energy management. This figure validates the training strategy and data preparation process. Overall, it provides visual confirmation of the LSTM’s performance.

This figure illustrates the difference between the true SOC and the estimated SOC from the LSTM network. The error remains relatively small throughout the test duration, indicating high estimation accuracy. Peaks in error correspond to periods of rapid load changes, reflecting the network’s transient response limitations. The figure helps quantify the reliability and precision of the proposed SOC estimation method. It highlights the robustness of the machine learning approach under dynamic and noisy conditions. Observing SOC error trends allows identification of potential improvements in network design or training. The figure is essential for calculating performance metrics like RMSE. It also shows the temporal distribution of estimation deviations. Small errors confirm the suitability of the method for real-time BMS applications.

This figure compares the true SOH obtained from the simulation with the estimated SOH derived using a capacity-based approach. The sliding window method calculates effective capacity based on SOC changes and cumulative current. The estimated SOH closely matches the true SOH trend, indicating successful degradation tracking. Minor fluctuations occur due to noise in SOC estimation and current integration. The figure highlights the ability of the method to capture long-term battery aging. Accurate SOH estimation enables predictive maintenance and lifespan optimization. It complements SOC estimation by providing health-aware insights. The figure demonstrates that capacity-based estimation combined with ML-derived SOC is effective. It shows gradual degradation over time, reflecting realistic battery aging.

This figure illustrates the continuous decrease in battery SOH due to cumulative current usage. The curve reflects the simulated capacity fade based on the defined degradation model. It demonstrates the slow, progressive nature of lithium-ion battery aging. This visualization provides context for the SOC and SOH estimation results. Observing capacity degradation helps validate the effectiveness of the estimation algorithms. It also highlights the need for long-term health monitoring in smart BMS. The figure serves as a reference for comparing estimated SOH values. It emphasizes the correlation between usage patterns and battery aging. The degradation trend informs maintenance scheduling and system reliability. Overall, it shows realistic battery behavior over extended operation.

You can download the Project files here: Download files now. (You must be logged in).

This figure shows the relationship between battery SOC and terminal voltage under dynamic load conditions. As SOC decreases, the voltage generally drops, with deviations caused by transient RC effects and load variations. The scatter plot reveals the nonlinear nature of this relationship, which complicates direct SOC estimation from voltage alone. It justifies the use of machine learning for capturing complex temporal dependencies. The figure helps visualize how voltage responds to both short-term and long-term battery dynamics. It also provides insight into periods of high stress or near-full/empty states. Observing the SOC-voltage pattern is useful for validating model predictions. The figure highlights the importance of combining multiple features for accurate estimation. Overall, it confirms that SOC cannot be reliably inferred from voltage without advanced modeling.

The figure shows the training progress of the LSTM network for SOC estimation. The top plot displays the RMSE versus iteration, where the error rapidly decreases in the initial epochs and stabilizes near zero, indicating effective learning. The bottom plot shows the training loss, which similarly drops quickly and reaches a low steady value, confirming convergence of the network. The training completed 60 epochs with a total of 2340 iterations, using a constant learning rate of 0.005 on a single CPU. Overall, the plots demonstrate successful network training with minimal residual error, validating the model’s capability to accurately estimate SOC from the input data.

6. Results and Discussion

The simulation results demonstrate the effectiveness of the proposed machine learning–based smart battery management system under dynamic load conditions.

Table 4: Performance Metrics

The applied current profile successfully excites both short-term and long-term battery behaviors, providing realistic input for state estimation [20]. The terminal voltage closely follows the expected trends predicted by the RC model, with voltage drops corresponding to high current demands and rises during low load intervals. The LSTM network accurately estimates SOC, with the predicted trajectory closely matching the true SOC over the test period. Minor deviations occur during rapid load changes, highlighting the transient response limits of the network [21]. The SOC estimation error remains low, and the calculated RMSE confirms high precision, validating the robustness of the approach. SOH estimation based on capacity modeling effectively tracks battery degradation, showing close agreement with the simulated true SOH [22]. The sliding window method allows the capture of cumulative aging effects while compensating for transient fluctuations in SOC. Comparison of estimated and true SOH illustrates the potential for predictive maintenance and long-term battery health monitoring. The scatter plot of SOC versus terminal voltage highlights the nonlinear relationship between these variables, emphasizing the limitations of voltage-based SOC estimation alone. The results demonstrate that combining LSTM-based SOC estimation with capacity-based SOH modeling provides a comprehensive solution for smart battery management. This hybrid approach ensures accurate short-term state monitoring while also addressing long-term aging trends. The methodology is robust to noise and variability in the current profile, ensuring generalization to real-world conditions. Performance metrics indicate that the integrated system can reliably support energy management and operational safety. Observed capacity degradation trends are consistent with expected lithium-ion battery aging patterns. The approach enables dynamic adaptation to changing load conditions without requiring explicit parameter recalibration [23]. Overall, the results confirm that machine learning can enhance the accuracy and reliability of BMS. The proposed system is suitable for applications in electric vehicles and stationary energy storage. These findings highlight the importance of data-driven methods for intelligent energy management and battery longevity optimization.

7. Conclusion

The study presents a machine learning–based smart battery management system capable of accurately estimating both SOC and SOH under dynamic load conditions. By integrating an LSTM network for SOC prediction with a capacity-based SOH estimation approach, the system captures short-term battery dynamics and long-term degradation trends [24]. Simulation results demonstrate that the estimated SOC closely follows the true SOC, with low error and high robustness under varying load profiles. SOH estimation effectively tracks capacity fade, enabling predictive maintenance and health-aware operation. The hybrid framework overcomes limitations of conventional model-based methods, such as sensitivity to parameter variations and noise. The methodology is adaptable to real-world conditions, including fluctuating currents and measurement uncertainties. Performance metrics, including RMSE, confirm the accuracy and reliability of the proposed system. Graphical analyses validate the effectiveness of the LSTM network and degradation modeling [25]. This approach enhances safety, efficiency, and lifespan of lithium-ion batteries in practical applications. Overall, the results highlight the potential of machine learning–driven BMS for intelligent energy storage management.

8. References

[1] X. Hu, S. Li, and Z. Peng, “A comparative study of equivalent circuit models for Li-ion batteries,” Journal of Power Sources, vol. 198, pp. 359-367, 2012.

[2] M. A. Hannan, M. S. H. Lipu, A. Hussain, and A. Mohamed, “A review of lithium-ion battery state of charge estimation and management system,” Renewable and Sustainable Energy Reviews, vol. 78, pp. 834-854, 2017.

[3] J. B. Gonder, T. Markel, M. Simpson, and M. Thornton, “Battery pack state of charge estimation,” National Renewable Energy Laboratory, 2007.

[4] Y. Li, K. Liu, A. M. Foley, A. K. Nandi, and J. Zhang, “A review of lithium-ion battery state of charge estimation methods,” Journal of Power Sources, vol. 448, p. 227401, 2020.

[5] S. S. Haykin, “Neural networks and learning machines,” Prentice Hall, 2009.

[6] J. Schmidhuber, “Deep learning in neural networks: An overview,” Neural Networks, vol. 61, pp. 85-117, 2015.

[7] F. Zhang, Z. Shi, and Y. Wang, “Lithium-ion battery state of charge estimation using a long short-term memory network,” Journal of Energy Storage, vol. 28, p. 101223, 2020.

[8] Y. Li, K. Liu, A. M. Foley, A. K. Nandi, and J. Zhang, “A review of lithium-ion battery state of charge estimation methods,” Journal of Power Sources, vol. 448, p. 227401, 2020.

[9] X. Hu, S. Li, and Z. Peng, “A comparative study of equivalent circuit models for Li-ion batteries,” Journal of Power Sources, vol. 198, pp. 359-367, 2012.

[10] M. A. Hannan, M. S. H. Lipu, A. Hussain, and A. Mohamed, “A review of lithium-ion battery state of charge estimation and management system,” Renewable and Sustainable Energy Reviews, vol. 78, pp. 834-854, 2017.

[11] J. B. Gonder, T. Markel, M. Simpson, and M. Thornton, “Battery pack state of charge estimation,” National Renewable Energy Laboratory, 2007.

[12] Y. Li, K. Liu, A. M. Foley, A. K. Nandi, and J. Zhang, “A review of lithium-ion battery state of charge estimation methods,” Journal of Power Sources, vol. 448, p. 227401, 2020.

[13] S. S. Haykin, “Neural networks and learning machines,” Prentice Hall, 2009.

[14] J. Schmidhuber, “Deep learning in neural networks: An overview,” Neural Networks, vol. 61, pp. 85-117, 2015.

[15] F. Zhang, Z. Shi, and Y. Wang, “Lithium-ion battery state of charge estimation using a long short-term memory network,” Journal of Energy Storage, vol. 28, p. 101223, 2020.

[16] Y. Li, K. Liu, A. M. Foley, A. K. Nandi, and J. Zhang, “A review of lithium-ion battery state of charge estimation methods,” Journal of Power Sources, vol. 448, p. 227401, 2020.

[17] X. Hu, S. Li, and Z. Peng, “A comparative study of equivalent circuit models for Li-ion batteries,” Journal of Power Sources, vol. 198, pp. 359-367, 2012.

[18] M. A. Hannan, M. S. H. Lipu, A. Hussain, and A. Mohamed, “A review of lithium-ion battery state of charge estimation and management system,” Renewable and Sustainable Energy Reviews, vol. 78, pp. 834-854, 2017.

[19] J. B. Gonder, T. Markel, M. Simpson, and M. Thornton, “Battery pack state of charge estimation,” National Renewable Energy Laboratory, 2007.

[20] Y. Li, K. Liu, A. M. Foley, A. K. Nandi, and J. Zhang, “A review of lithium-ion battery state of charge estimation methods,” Journal of Power Sources, vol. 448, p. 227401, 2020.

[21] S. S. Haykin, “Neural networks and learning machines,” Prentice Hall, 2009.

[22] J. Schmidhuber, “Deep learning in neural networks: An overview,” Neural Networks, vol. 61, pp. 85-117, 2015.

[23] F. Zhang, Z. Shi, and Y. Wang, “Lithium-ion battery state of charge estimation using a long short-term memory network,” Journal of Energy Storage, vol. 28, p. 101223, 2020.

[24] Y. Li, K. Liu, A. M. Foley, A. K. Nandi, and J. Zhang, “A review of lithium-ion battery state of charge estimation methods,” Journal of Power Sources, vol. 448, p. 227401, 2020.

[25] X. Hu, S. Li, and Z. Peng, “A comparative study of equivalent circuit models for Li-ion batteries,” Journal of Power Sources, vol. 198, pp. 359-367, 2012.

You can download the Project files here: Download files now. (You must be logged in).

Do you need help with A Hybrid Machine Learning Framework for Accurate SOC and SOH Estimation in Lithium-Ion Batteries in MATLAB? Don’t hesitate to contact our Tutors to receive professional and reliable guidance.