Multi-objective optimization estimation of state of health for lithium-ion battery based on constant current charging profile

The capacity decline, i.e., the state of health trajectory of lithium-ion battery is strongly nonlinear and volatile. Currently, most studies on the state of health assessment focus more on improving the estimation accuracy, rarely taking result stability into consideration. In this paper, a method adopting the multi-objective optimization extreme learning machine for estimating the state of health of lithium-ion battery based on the constant-current charging curve of the battery is proposed to improve both the estimation accuracy and stability. Firstly, apart from isovoltage rise charging time, a logit polynomial fitting model is taken to fit the constant-current charging curve to extract the features which highly associated with battery capacity level and proven by the Spearman coefficient. Then, the MOWOA-ELM prediction model is built for state of health estimation where the multi-objective whale search algorithm improved by non-dominated sorting and congestion calculation is used to optimize the hyperparameters of the extreme learning machine. Finally, the experimental and comparative results show that the root mean square error and the standard deviation of the state of health assessment results are only 0.43 % and 0.28 % respectively, demonstrating the feasibility and validity of the framework built in this paper.