A New Optimization Algorithm for Parameters Identification of Electric Vehicles' Battery

This study deals with parameter identification of behavioral model of the electric vehicle's (EV) battery, which can be cast as a difficult optimization problem. This necessitates the employment of a powerful and global optimization algorithm to ensure the reliability of the results. In this study, a newly developed optimization technique referred to as evolutionary-particle swarm optimization (E-PSO) is implemented. A statistical analysis is conducted, and the proposed algorithm is compared with other widespread metaheuristic algorithms in terms of convergence and simulation time. To do so, first, the current of the battery is determined using a typical EV model and a standard driving cycle. Then, experimental tests are conducted on Lithium Polymer off the shelf cell to calculate the actual terminal voltage. Finally, this actual data is used in an optimization frame to calculate the parameters of the model by which the behavioral model and the real battery are in the closest agreement. The results show that the E-PSO algorithm outperforms other metaheuristic optimization algorithms in terms of finding better solution in a lower convergence time. It is also demonstrated that the solution obtained by E-PSO provides a more accurate estimation of the actual battery.