Health indicator evaluation for battery pack inconsistency and SOH estimation based on LSTM

State-of-health (SOH) estimation of lithiumion battery is one of the significant factors in improving battery performance. For real applications, it is however difficult to estimate battery SOH due to complex problem, which involves non-linear characteristics of battery degradation and inconsistency among constituent cells during battery application is operating. Thus, it is an important tack to select health indicators (HI) that can reflect not only battery degradation but also inconsistency changes at the same time. In this study, Pearson correlation analysis is applied in this study to manifest evaluate the effectiveness of HIs over cycle. To further evaluate the robustness of HIs, SOH estimation of the battery pack is conducted based on long short term memory (LSTM). And the results show that selected features have good performance to estimate battery SOH.