An Optimized Multiple-Weighted Adaptive Genetic-Kalman Hybrid Al-Gorithm for Online State of Charge Estimation in Lithium-Ion Batteries

As an indispensable aspect of the battery management system, accurate lithium battery state of charge (SOC) estimation has attracted wide attention. This study proposes an innovative adaptive genetic algorithm online intermittent parameter identification method to improve the model parameter identification accuracy based on the second-order RC equivalent circuit as the battery model. Addressing the limitations of offline identification and the fixed crossover and mutation probabilities in the traditional genetic algorithm enhances the algorithm’s efficiency, thereby overcoming the constraints associated with offline identification. In addition, an improved multiple-weighted adaptive extended Kalman filter algorithm is proposed to enhance SOC estimation accuracy further. This approach aims to address the issue of an ineffective proportional weighting of historical new interest, which can lead to a potential slowdown in convergence and an increase in error when historical new interest exhibits significant fluctuations. Ultimately, the SOC estimation results based on the above two algorithms are exhaustively compared and analyzed under HPPC and BBDST working conditions, which show that the simulated voltage error decreases up to 44.6%, and none of the SOC estimation errors exceed 1.2%. It provides a theoretical basis for the practical application of the battery management system system.