SOH estimation of lithium-ion battery under complex operating conditions based on BP neural network

The precise state of health (SOH) estimation and monitoring of lithium-ion batteries are extremely important for future BMS (intelligent battery management system). However, there is a lack of SOH estimation methods for complex on-board conditions in the existing research results. This research suggests a battery SOH estimate technique based on the battery's CCCV (constant voltage and constant current) charging curve and BP neural network in order to solve this issue. Firstly, multiple health indicators are extracted and screened by analyzing the correlation between battery CCCV charging curve characteristics and battery SOH. Then, the lithium-ion battery data under complex operating conditions are corrected by the empirical capacity model, and the SOH estimation model of lithium-ion battery is constructed by using the BP neural network. In conclusion, this study trains and validates the model using battery aging experimental data from NASA (National Aeronautics and Space Administration). The results demonstrate that the model's error is less than 5%, confirming the model's efficacy.