Divyakumar Bhavsar
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Divyakumar Bhavsar has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Kanupriya Mittal
Anna University
2 shared publications Mayank Mittal
Department of Mechanical Engineering Indian Institute of Technology Madras Chennai India
2 shared publicationsPratik Mondal
Department of Mechanical Engineering, Indian Institute of Technology, Madras, Chennai, India
2 shared publicationsKanupriya Mittal
VERDOJAH TECHNOLOGIES PRIVATE LIMITED, Chennai, India
1 shared publicationLeelakumar Murugesan
Department of Electrical and Electronics Engineering National Institute of Technology Tiruchirappalli Tiruchirappalli India
1 shared publication- Mayank Mittal
Indian Institute of Technology Madras, Chennai, India
1 shared publication
Latest Publications
- 4 September 2026
- Published by Wiley in Energy Storage
- Vol. 8 (6)
Lithium‐Ion Battery Parameter Estimation With Adaptive Honey Badger Optimization and State‐of‐Charge Prediction Using the Extended Kalman Filter
Accurate parameter identification and terminal‐voltage estimation are essential for precise state‐of‐charge estimation and effective control of lithium‐ion battery management systems. Conventional optimization methods, such as Particle Swarm Optimization (PSO) and Honey Badger Optimization (HBO), are prone to local optima under certain nonlinear operating conditions and are less effective at identifying parameter values that minimize the root mean square error (RMSE) in nonlinear battery models at specific temperatures. Hence, this study introduces the Adaptive Honey Badger Optimization (AHBO) algorithm by incorporating Sine‐chaotic population, adaptive exploration, adaptive exploitation, diversity preservation, and conditional recomputation mechanisms, thereby maintaining an effective balance between global exploration and local exploitation. It has been achieved through adaptively regulated search dynamics and a diversity‐dependent exponential‐decay mechanism. The proposed AHBO method, with static OCV–SoC values, is used for parameter identification and battery voltage estimation in the equivalent circuit models of the One‐RCH and Two‐2RCH configurations. The results demonstrate that AHBO achieves lower terminal‐voltage estimation RMSE values, ranging from 7.85 to 16.64 mV for the Two‐RCH configuration over the temperature range of 45°C to −15°C, compared with PSO and HBO. In addition, the proposed method reduces the convergence time by approximately 70%–75% and enhances the dynamic terminal‐voltage response, particularly at low temperatures. After parameter identification, a comparative study of state‐of‐charge estimation was conducted using the Extended Kalman Filter (EKF) and the Gated Recurrent Unit (GRU) methods. The EKF exhibited a physically consistent model‐based SoC response with RMSE values of 0.87% at 45°C, 1.18% at 15°C, and 1.48% at −15°C. However, the GRU demonstrated better estimation accuracy at the intermediate temperature of 15°C (RMSE: 0.73%) and comparable accuracy at −15°C (RMSE: 1.40%), while showing reduced accuracy at 45°C (RMSE: 1.36%).
- 4 September 2026
- Published by Wiley in Energy Storage
- Vol. 8 (6)
Lithium‐Ion Battery Parameter Estimation With Adaptive Honey Badger Optimization and State‐of‐Charge Prediction Using the Extended Kalman Filter
Accurate parameter identification and terminal‐voltage estimation are essential for precise state‐of‐charge estimation and effective control of lithium‐ion battery management systems. Conventional optimization methods, such as Particle Swarm Optimization (PSO) and Honey Badger Optimization (HBO), are prone to local optima under certain nonlinear operating conditions and are less effective at identifying parameter values that minimize the root mean square error (RMSE) in nonlinear battery models at specific temperatures. Hence, this study introduces the Adaptive Honey Badger Optimization (AHBO) algorithm by incorporating Sine‐chaotic population, adaptive exploration, adaptive exploitation, diversity preservation, and conditional recomputation mechanisms, thereby maintaining an effective balance between global exploration and local exploitation. It has been achieved through adaptively regulated search dynamics and a diversity‐dependent exponential‐decay mechanism. The proposed AHBO method, with static OCV–SoC values, is used for parameter identification and battery voltage estimation in the equivalent circuit models of the One‐RCH and Two‐2RCH configurations. The results demonstrate that AHBO achieves lower terminal‐voltage estimation RMSE values, ranging from 7.85 to 16.64 mV for the Two‐RCH configuration over the temperature range of 45°C to −15°C, compared with PSO and HBO. In addition, the proposed method reduces the convergence time by approximately 70%–75% and enhances the dynamic terminal‐voltage response, particularly at low temperatures. After parameter identification, a comparative study of state‐of‐charge estimation was conducted using the Extended Kalman Filter (EKF) and the Gated Recurrent Unit (GRU) methods. The EKF exhibited a physically consistent model‐based SoC response with RMSE values of 0.87% at 45°C, 1.18% at 15°C, and 1.48% at −15°C. However, the GRU demonstrated better estimation accuracy at the intermediate temperature of 15°C (RMSE: 0.73%) and comparable accuracy at −15°C (RMSE: 1.40%), while showing reduced accuracy at 45°C (RMSE: 1.36%).
- 6 April 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering
State-of-charge estimation of two-wheeler electric vehicle battery pack by long-short term memory model trained using cell-level data
The rising global pollution and the urgent shift towards sustainable transportation have accelerated the adoption of electric vehicles (EVs). EV efficiency and performance is majorly dependent on the lithium-ion battery pack, which requires accurate State of Charge (SoC) estimation to ensure effective battery management and longevity. Traditional methods like Coulomb counting and model-based approaches often struggle with cumulative errors, parameter sensitivity, and modelling inaccuracies. This study explores the applicability of deep-learning based model for estimation of pack-level SoC when trained using cell-level data, even when battery chemistry and ambient temperature are unknown. Long short-term memory (LSTM) networks are initially trained on individual cell-level datasets and tested on pack-level data to assess the applicability. In order to investigate the feasibility of a generalized model, LSTM model, further, is trained on multiple cell-level datasets consisting of two different chemistries and different temperatures and tested on pack-level dataset. The findings indicate that models trained with individual or multiple cell-level datasets can effectively estimate pack-level SoC. Results confirm the model’s robustness in estimating SoC without prior knowledge of battery chemistry.
- 6 April 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering
State-of-charge estimation of two-wheeler electric vehicle battery pack by long-short term memory model trained using cell-level data
The rising global pollution and the urgent shift towards sustainable transportation have accelerated the adoption of electric vehicles (EVs). EV efficiency and performance is majorly dependent on the lithium-ion battery pack, which requires accurate State of Charge (SoC) estimation to ensure effective battery management and longevity. Traditional methods like Coulomb counting and model-based approaches often struggle with cumulative errors, parameter sensitivity, and modelling inaccuracies. This study explores the applicability of deep-learning based model for estimation of pack-level SoC when trained using cell-level data, even when battery chemistry and ambient temperature are unknown. Long short-term memory (LSTM) networks are initially trained on individual cell-level datasets and tested on pack-level data to assess the applicability. In order to investigate the feasibility of a generalized model, LSTM model, further, is trained on multiple cell-level datasets consisting of two different chemistries and different temperatures and tested on pack-level dataset. The findings indicate that models trained with individual or multiple cell-level datasets can effectively estimate pack-level SoC. Results confirm the model’s robustness in estimating SoC without prior knowledge of battery chemistry.
- Open Access
- 3 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 78773-78786
Estimating State-of-Charge in Lithium-Ion Batteries Through Deep Learning Techniques: A Comparative Evaluation
Accurate estimation of the state of charge (SoC) is crucial for optimizing battery performance, battery health estimation, and ensuring reliable operation. In recent years, deep learning techniques have shown promising results in capturing complex non-linear relationships between input features and estimated the SoC as they do not rely on battery models. However, a detailed comparison of these techniques is not available. Therefore, this paper presents a comprehensive comparison of SoC estimation using ANN (artificial neural network), LSTM (long short-term memory), BiLSTM (bidirectional LSTM), and GRU (gated recurrent unit) models applied to lithium-ion batteries. These models are trained using five distinct drive cycles and tested against four additional drive cycles from a publicly available dataset. The effectiveness is evaluated with RMSE (root mean squared error) and R2 score (coefficient of determination). The results provide insights into the applicability of ANN, LSTM, BiLSTM, and GRU for SoC estimation in lithium-ion battery systems, highlighting the advantages and limitations of each architecture. Additionally, the neuron effect is considered to examine the impact of the number of neurons in the hidden layer on the prediction of SoC. Comparing all models, it is observed that the BiLSTM network shows promising results, showing RMSE of 0.0097 and R2 value of 0.9987 when evaluated on one of the testing datasets, surpassing the performance of the other models. Further, the addition of temperature as a feature and the effect of external noise on the SoC estimation using the BiLSTM network are analyzed, highlighting the significance of accurate SoC estimation.
- 3 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 78773-78786
Open AccessEstimating State-of-Charge in Lithium-Ion Batteries Through Deep Learning Techniques: A Comparative Evaluation
Accurate estimation of the state of charge (SoC) is crucial for optimizing battery performance, battery health estimation, and ensuring reliable operation. In recent years, deep learning techniques have shown promising results in capturing complex non-linear relationships between input features and estimated the SoC as they do not rely on battery models. However, a detailed comparison of these techniques is not available. Therefore, this paper presents a comprehensive comparison of SoC estimation using ANN (artificial neural network), LSTM (long short-term memory), BiLSTM (bidirectional LSTM), and GRU (gated recurrent unit) models applied to lithium-ion batteries. These models are trained using five distinct drive cycles and tested against four additional drive cycles from a publicly available dataset. The effectiveness is evaluated with RMSE (root mean squared error) and R2 score (coefficient of determination). The results provide insights into the applicability of ANN, LSTM, BiLSTM, and GRU for SoC estimation in lithium-ion battery systems, highlighting the advantages and limitations of each architecture. Additionally, the neuron effect is considered to examine the impact of the number of neurons in the hidden layer on the prediction of SoC. Comparing all models, it is observed that the BiLSTM network shows promising results, showing RMSE of 0.0097 and R2 value of 0.9987 when evaluated on one of the testing datasets, surpassing the performance of the other models. Further, the addition of temperature as a feature and the effect of external noise on the SoC estimation using the BiLSTM network are analyzed, highlighting the significance of accurate SoC estimation.