RUL Prediction for Lithium-ion Batteries Using Combination Forecasting based on SVR and LSTM

Authors:

Abstract
Forecasting the remaining Useful Life (RUL) of Lithium-ion battery has been a major concern in the energy industry. The poor learning ability of local features and insufficient learning ability of large amounts of data are some problems existing. In order to solve the existing problems, a data-driven prediction method combining support vector regression (SVR) and long short-term memory (LSTM) is proposed. Taking standard deviation as the criterion, the method continuously optimizes the constrained multivariate function through sequence quadratic program, and iterates to obtain the minimum error value to calculate the optimal weight corresponding to each model, then the prediction accuracy is improved by combination. The experimental results show that the root mean square error of the combination forecasting is around 1% when the number of cycles is specified, which provides a good reference for improving the performance of the battery management system.
Funding Information
  • Natural Science Foundation of Fujian Province

This publication has 11 references indexed in Scilit: