Gradient-based optimization for parameter identification of lithium-ion battery model for electric vehicles

Defining proper parameters for lithium-ion battery models is a challenge for several applications, including automobiles powered by electricity. Conventional parameter identification methods rely on human tweaking or experimentation and failure procedures, and this may be time-consuming and produce unsatisfactory results. In the past few years, metaheuristic optimization approaches have developed as useful instruments for identifying and determining appropriate settings for parameters. This study proposes an effective parameter determination approach for applications on electric vehicles using an evolutionary optimization methodology and a Shepherd model. The identification approach based on the gradient-based optimizer worked exceptionally well in determining the battery's equivalent circuit parameters. The root means square error between the battery's real data and its model is the target to be minimized. The findings were contrasted to those from various algorithms, such as whale optimization algorithm, multi-verse optimizer, sine cosine algorithm, arithmetic optimization algorithm, particle swarm optimization, red kite optimization algorithm, tree-seed algorithm, and white shark optimizer. As a result, the recommended identification approach outperformed as the overall error in voltage was decreased to 4.2377 x 10(-3), and the RMSE difference between the predicted value and the actual data was 8.64 x 10(-3).