基于SSA-SVR和LSTM相结合的混合模型预测锂电池的剩余寿命

Predicting the Remaining Life of Lithium Batteries Based on a Hybrid Model Combining SSA-SVR and LSTM

锂电池的SOH和RUL可以判断电池管理系统故障的几率。文章提出一种预测SOH和RUL的混合模型。首先利用改进的带有自适应噪声的互补集合经验模态分解算法(ICEEMDAN)分解容量信号,然后分别利用SVR算法、LSTM对高频、低频信号进行预测,同时引入SSA优化SVR参数以提高精度,最后将各分量预测信号重组作为最终的预测结果。仿真结果表明,在不同数据集上各项预测评估指标均小于1%,该混合预测模型具有稳定性好、精度高和鲁棒性强等优点,适用于预测电池SOH和RUL。

The SOH and RUL of Li-ion batteries can determine the chance of battery management system failure.In this paper,a hybrid model for predicting SOH and RUL is proposed.Firstly,the capacity signal is decomposed using the improved complementary ensemble empirical modal decomposition algorithm with adaptive noise(ICEEMDAN),and then the high-frequency and low-frequency signals are predicted using the SVR algorithm and LSTM,respectively.And at the same time,the SSA is introduced to optimize the SVR parameters to improve the accuracy,and fi nally,the predicted signals of each component are reorganized as the fi nal prediction results.The simulation results show that all the prediction evaluation indexes are less than 1%on diff erent datasets,and the hybrid prediction model has the advantages of good stability,high accuracy and robustness,which is suitable for predicting the SOH and RUL of batteries.