基于改进相关向量机的锂电池剩余使用寿命预测

Remaining useful life prediction of Li-ion batteries based on improved relevance vector machine

精确预测锂离子电池剩余使用寿命对于保障设备安全运行十分重要。但电池寿命预测中存在诸如数据噪声和容量再生等不确定性来源,这将导致预测精度大幅下降。为解决这一问题,使用变分模态分解方法对从充电和容量数据中提取的健康因子进行滤波分解,并利用贝叶斯优化方法对相关参数进行优化,提出一种基于多核相关向量机的锂离子电池剩余使用寿命预测模型。利用美国国家航空航天局(NASA)和Oxford电池数据集对所提出的模型进行验证,研究结果表明:所提出的基于变分模态分解和贝叶斯优化的多核相关向量机(VMD-BAYES-HRVM)方法的预测性能不受预测起始点和截止电压的影响,预测结果准确性更高,95%置信区间的跨度更小,证明了所提出方法的有效性。

The accurate prediction of the remaining life of lithium-ion batteries is important to ensure safe operation of equipment.However,the sources of uncertainty in battery life prediction,such as data noise and capacity regeneration,can lead to significant degradation of prediction accuracy.To solve the problem,the variational mode decomposition was used to filter and decompose the health factors extracted from charging and capacity data,the relevant parameters were optimized by using the Bayesian optimization method,and a Li-ion battery remaining life prediction model based on the hybrid kernel relevance vector machine was proposed.The proposed model was validated by using the NASA and Oxford battery datasets.The results show that the prediction performance of the proposed VMD-BAYES-HRVM method is not affected by the prediction starting point and cut-off voltage,and the prediction results are more accurate and have a smaller span of 95%confidence interval,which proves the effectiveness of the proposed method.