基于DAE-HTPF的新能源汽车电池剩余寿命预测

针对反映锂电池寿命的趋势性特征自学习与电池剩余寿命预测问题，提出了基于降噪自编码器（denoising auto-encoder，DAE）与混合趋势粒子滤波（hybrid trend particle filter，HTPF）的电池剩余寿命预测方法。利用电池使用前期的信号特征训练DAE，然后将使用中后期的电池信号特征输入DAE中，并提取重构误差。另外，利用HTPF方法对电池生命周期内的信号特征进行分析，建立自适应状态方程。分析结果表明，该方法能有效地对锂电池的性能退化趋势性特征进行自提取，从而有效地减少人为因素的干扰，同时相比于传统粒子滤波（particle filter，PF），HTPF对电池剩余寿命预测精度更高。

Remaining Life Prediction of New Energy Vehicle Battery Based on DAE-HTPF

A method of battery life prediction based on Denoising Auto-encoder(DAE) and Hybrid Trend Particle Filter(HTPF) was proposed to analyze the trend feature of self-learning for lithium battery and its remaining life prediction. The DAE was first trained by the signal characteristics of the battery in its early stage, then input with the signal characteristics of it in middle and late stages, and the reconstruction error data was extracted. In addition, the HTPF method was used to analyze the signal characteristics in the battery life cycle, and an adaptive state equation was established. The analysis results show that this method can effectively extract the degradation trend characteristics of lithium battery, thus effectively reducing the interference of human factors. Compared with traditional particle filter(PF), HTPF is more accurate in predicting battery remaining life.