State of health estimation joint improved grey wolf optimization algorithm and LSTM using partial discharging health features for lithium-ion batteries

Accurate estimation of the state of health (SOH) of lithium-ion batteries is crucial for the safety and operation of electric vehicles. The accuracy and efficiency of SOH estimation are degraded by existing data-driven methods that depend on the empirically selecting hyperparameters and time-consuming aging data. In this paper, a method joint improved grey wolf optimization (IGWO) algorithm and long short-term memory (LSTM) is developed to estimate the SOH using partial discharging health features (HFs). A dropout technique is applied to overcome the overfitting issue of the LSTM for SOH estimation. An IGWO algorithm is presented to address the challenges of the GWO algorithm, such as its tendency to fall into local optimization used with an LSTM, for accurately obtaining the optimal hyperparameters of LSTM. To reduce the consuming time of aging data, the LSTM model joint the IGWO is developed to estimate the SOH using partial discharging HFs. Compared to using five HFs, the experimental results demonstrate that the SOH can be estimated accurately by the developed method using two HFs in shorter consuming time with the mean absolute error, root mean square error, mean absolute percentage error, and average values of all of them within 1 %.