基于KPCA特征集成算法的SOFC 系统多故障识别

Multi-fault Identification of SOFC System Based on KPCA Feature Integration Algorithm

针对固体氧化物燃料电池系统多模式、非线性及高维等特点,提出基于数据驱动的模式识别方法。首先用核主成分分析特征集成算法提取故障特征,然后在特征空间中使用多项式逻辑斯谛回归算法进行故障诊断。实验结果表明:核主成分分析特征集成算法可以全面提取出故障特征,能够大幅提高后续分类器故障的识别效果。

Considering multi-mode, nonlinearity and high-dimensional characteristics of SOFC system, a data-driven mode recognition method was proposed. Firstly, the kernel principal component analysis feature integration algorithm was applied to extract fault features, and then the multinomial logistic regression algorithm was used in the feature space to diagnose the faults. The experimental results show that, the kernel principal component analysis feature integration algorithm can effectively extract fault features and greatly improve the fault recognition effect of subsequent classifiers.