Abstract:

Aiming at the problem that the existing online life prediction of lithium?ion batteries based on the correlation vector machine has a single consideration factor, which results in unsatisfactory prediction accuracy, a method based on principal component analysis (PCA) for weighted construction of characteristic factor variables was proposed. In this method, a variety of characteristic factor variables are taken as the research object to find the matrix of the score vector after the linear transformation. The feature coverage degree of different score vectors to the original variable data matrix is analyzed, and the corresponding feature vectors are constructed by weighted fusion. Using the vector as input, a prediction model is established by the relevance vector machine and the online prediction of lithium?ion battery life is performed, and the prediction results are finally obtained. International public battery data was used as the research object, and MATLAB experiments were used to verify that the method has the feasibility of multivariate prediction of battery life, and the prediction effect is better.

Key words: Principal component analysis, Relevance vector machine, Linear correlation, Life prediction

摘要：

针对已有基于相关向量机对锂离子电池进行在线寿命预测因考虑因素单一而导致预测精度不理想这一问题，提出了一种基于主元分析（PCA）的特征因素变量加权建构的方法。该方法首先将多种特征因素变量作为研究对象，找到其线性变换后的得分向量所构矩阵；分析其不同得分向量对原变量数据矩阵特征覆盖程度，进一步加权构建融合得到相应特征向量。将所得向量作为输入，经相关向量机建立预测模型并进行锂离子电池寿命在线预测，最终得到预测结果。采用国际公用电池数据作为研究对象，通过MATLAB软件验证了有多变量预测电池寿命的可行性，结果表明预测效果较好。

关键词: 主元分析, 相关向量机, 线性相关性, 寿命预测