关键词: 锂离子电池, 经验模态分解, 健康状态预测, 容量再生, 融合模型

Abstract: The prediction of battery state of health(SOH) is a key factor to ensure the reliability and safety of electronic system operation. In order to accurately predict the overall degradation trend and local capacity regeneration of lithium-ion battery SOH, a lithium-ion battery SOH prediction method combining empirical mode decomposition(EMD), gated recurrent unit(GRU) and differential autoregressive integrated moving average model(ARIMA) is proposed. First, the original SOH sequence of the battery is decomposed at multiple scales using EMD, and the high and low frequency demarcation points are found by calculating the continuous mean square error of the decomposed subsequences; then, GRU is used to predict high-frequency subsequences with strong data fluctuations, and ARIMA is used to predict the remaining low-frequency subsequences and residuals; finally, the prediction results of each subsequence are superimposed to obtain the final prediction result. The experimental results show that, compared with the prediction methods in other literatures, the fusion model based on empirical mode decomposition has higher prediction accuracy and can better capture the overall degradation trend and local capacity regeneration characteristics of battery SOH.

Key words: lithium-ion battery, empirical mode decomposition, state of health prediction, capacity regeneration, fusion model