State of Health Estimation for Lithium-ion Batteries Based on Fusion of Autoregressive Moving Average Model and Elman Neural Network

Abstract

This paper proposes a fusion model based on the autoregressive moving average (ARMA) model and Elman neural network (NN) to achieve accurate prediction for the state of health (SOH) of lithiumion batteries. First, the voltage and capacity degradation variation of the battery are acquired through the battery lifecycle data, and the health factor related to the battery aging is selected according to the variation of the voltage profile. Second, the empirical mode decomposition (EMD) is employed to process the capacity degradation data and eliminate the phenomenon of tiny capacity recovery, and multiple data sequences, as well as the related residue, are extracted, then the grey relational analysis (GRA) between sub-sequences and health factor are discussed. Furthermore, the ARMA model and Elman NN model are respectively built by training the subsequent time series data and residue data. Finally, all the individual predictions are combined to generate the estimated SOH sequences. The experimental validation is performed to manifest that the addressed fusion method performs the SOH prediction with satisfactory accuracy, compared with the single ARMA method and Elman NN model. INDEX TERMS Autoregressive moving average (ARMA), Elman neural network (NN), grey relational analysis (GRA), empirical mode decomposition (EMD), state of health (SOH).

Online State-of-Health Estimation for Second-Use Lithium-Ion Batteries Based on Weighted Least Squares Support Vector Machine

“…Li et al [27] selected the ratio of constant current (CC) during the charging phase and the time spent in the selected equidistant voltage interval as the healthy features. Chen et al [28] chose the CC mode duration as the input feature in their proposed method. Zheng et al [29] took the area under the CC mode, the time duration from 3.9 V to 4.2 V under the CC mode and the end time of the CC mode as the training data sets.…”

Joint Trajectory Prediction of Multi-Linkage Robot Based on Graph Convolutional Network

“…In our future work, the mean-subtraction trick [56] will be used to improve the phenomenon, which is that subtracting the mean value for each feature channel of the hidden layer. It is one of the classical time series methods [57].…”

State-of-Health Prediction for Lithium-Ion Batteries Based on a Novel Hybrid Approach

“…It is suitable for nonlinear small sample problems. Least squares support vector machine (LSSVM) transforms the problem into solving linear equations and the convergence speed is faster [37][38][39][40][41][42]. Next, a brief description of its principle follows.…”