Enhancing global search of honey badger algorithm for High-Accuracy Lithium-Ion battery modeling and SOC estimation

• Developed a modified honey badger algorithm with Brownian-motion search mechanism. • Improved global exploration and reduced local stagnation in high-dimensional optimization. • Achieved higher accuracy in lithium-ion battery parameter identification. • Achieved SOC estimation accuracy improvements of up to 53.564% with low voltage tracking error. Accurate modeling and reliable SOC estimation are central to battery management system design. Lithium-ion batteries show nonlinear behavior and create a multimodal search landscape that makes complex optimization challenging. The honey badger algorithm (HBA) converges rapidly in complex optimization problems but exhibits a strong exploitation bias that limits global exploration. To improve solution stability and search robustness, the modified honey badger algorithm (MHBA) with Brownian-motion is proposed to enhance global search capability. MHBA is first tested on 23 standard optimization functions to assess robustness and is then applied to lithium-ion battery parameter identification, where performance is compared with four advanced algorithms in terms of best fitness, mean fitness, standard deviation and search behavior. Subsequently, the MHBA-optimized parameters are incorporated into extended Kalman filter to examine the impact of modeling accuracy on SOC estimation under three temperature conditions (0 °C, 25 °C and 50 °C), across different discharge rates and initial SOC conditions. The results show that the RMSE values obtained using MHBA and HBA are 1.512 mV and 1.553 mV, respectively. In SOC estimation validation, the MHBA-based parameters improve estimation accuracy by up to 53.564 %. These results indicate that MHBA not only effectively enhances search capability but also provides a reliable foundation for the development of advanced battery management system.