An approach for state-of-health estimation based on VMD-LSTM

State-of-health (SOH) estimation is a critical factor to assess lithium-ion battery for safety and efficiency. However, it is challenging to achieve accurate estimation of SOH due to the extremely complex reaction inside. In order to improve the accuracy of SOH estimation, this paper proposed an improved data driven approach that combining LSTM model with VMD algorithm. The peak of incremental capacity curve (ICP) and the time interval of charging (TIOC) are selected as health indicators. Aiming at the regeneration phenomenon of the capacity, the SOH is decomposed into two components to estimate, respectively. The HIs are decomposed into two components with VMD algorithm, and the components of battery SOH are estimated on this basis. Then, the approach in this article is evaluated using two public data sets and compared the performance with other methods. The results show that our approach has an overall superior estimation performance in terms of both accuracy and applicability.