基于数据驱动的锂电池剩余使用寿命预测

Residual Life Prediction of Li-ion Batteries Based on Data Driving

基于数据驱动的思想,从电池历史数据中提取能反映电池衰退趋势的特征参数,并分析参数与电池寿命的相关性,完成特征参数的选取。其次基于选取的特征参数,对其进行数据预处理,得到最终的特征数据。最后基于时序预测的思想,建立长短期记忆神经网络的锂电池剩余使用寿命预测模型,从而实现电池剩余使用寿命的预测。研究结果表明:与传统的支持向量回归方法相比,基于长短期记忆神经网络的方法有效提高了预测准确性。

Based on the idea of data driving, having characteristic parameters which reflecting battery decline trend from the battery historical data extracted and the correlation between the parameters and the battery life analyzed as well as the selection of characteristic parameters completed were implemented, including having the selected feature parameters based to preprocess the data so as to obtain the final feature data, and having the idea of time series prediction and the long-short-term memory neural network based to establish lithium battery’s residual life prediction model so as to predict the residual life of the lithium battery. The results show that, compared with the traditional support vector regression method, the method based on the long short-term memory neural network can improve the prediction accuracy effectively.