PSO Assisted Support Vector Machine Algorithm Towards State of Charge Estimation of Li-ion Batteries

This research focuses on the State of Charge (SoC) estimation of Lithium-ion batteries employing the Support Vector Machine (SVM) algorithm. The five hyperparameters of SVM are optimally identified using Particle Swarm Optimization (PSO). Four publicly available driving cycles, namely US06, LA92, CSHVC, and HWFET are employed for the optimization and subsequent evaluation of the proposed method. The first 70% of each dataset is used for the optimization of hyperparameters and the remaining 30% is used for the testing of the algorithm. The SVM with optimized hyperparameters is then employed for SoC estimation of each driving cycle at different temperatures and the results are presented. It is observed that the performance indices, namely RMSE, MAE, and $\mathbf{R}^{2}$ values are far superior to the values in the available literature. Further, the percentage error in SoC is found to be below 2.5%, indicating improved performance of the proposed method.