An Improved Capacity-Loss Diagnostic Model Based on Long Short-Term Memory Network

Due to the capacity-loss of lithium-ion batteries is caused by the coupling effect of multiple mechanisms, the online capacity-loss diagnosis is still a challenge, and diagnosing the capacity-loss by using the theoretical model needs considerable time and cost. To solve the above problems, an improved model for online capacity-loss diagnosis based on long short-term memory neural networks (LSTM) is proposed. The network architecture of the model is designed based on the modeling process of the theoretical model. The powerful ability of feature extraction of LSTM network is utilized to solve the problem of model parameter identification, and the network architecture can reduce the dependence of the model on training data to some extent. The verification results show that, compared with the unidentified theoretical model, the proposed model improves the accuracy of capacity-loss diagnosis by 2% by training the model, and it has better adaptability to different batteries.

Automated summary

Due to the coupling effect of multiple mechanisms, the online capacity-loss diagnosis of lithium-ion batteries is still challenging and time-consuming. To address this issue, an improved model based on long short-term memory neural networks (LSTM) is proposed, which utilizes the powerful feature extraction ability of LSTM to identify model parameters and reduce dependence on training data. The verification results indicate that the proposed model improves the accuracy of capacity-loss diagnosis by 2% compared to the unidentified theoretical model, and exhibits better adaptability to different batteries.