Robust estimation of lithium-ion battery state of health based on electro-thermal features and machine learning

Accurately estimating the State of Health (SOH) of lithium-ion (Li-ion) batteries is crucial for preventing overcharging and over-discharging, thereby extending battery lifespan. This paper presents a novel method for SOH estimation in lithium-ion batteries by leveraging electro-thermal features, a backpropagation neural network (BPNN), and a particle swarm optimization (PSO) algorithm. First, three health-related features—constant current charging time (CCCT), relative constant current charging time (RCCCT), and maximum temperature during discharge (MTT)—are extracted as indicators of SOH. The correlation between these features and the SOH is validated using Grey Relational Analysis (GRA). Next, a BP neural network is utilized to model the nonlinear relationship between the extracted features and SOH. The PSO algorithm is then employed to optimize the parameters of the BP neural network, enhancing the accuracy of the SOH estimation. The proposed method, which combines electro-thermal features with the optimized BP neural network, is validated through three independent lithium-ion battery aging experiments. Experimental results demonstrate that the proposed approach achieves high estimation accuracy and exhibits strong generalization performance for SOH prediction.