论文部分内容阅读
针对功率变换器开关管的故障诊断和定位问题,提出了一种基于遗传算法优化的 M-ary 支持向量机故障分类方法。该方法以类内类间距离为判据,通过遗传算法来确定与故障本身有关的较优故障编码,然后建立优化后的 M-ary支持向量机模型来对功率变换器进行故障诊断和定位。仿真和物理实验结果表明,基于遗传算法优化后的M-ary支持向量机与标准M-ary支持向量机相比,具有更高的诊断精度;与常用的一对一支持向量机和一对多支持向量机相比,需要的子分类器数目少,诊断速度快。“,”Focusing on the issue of power converter fault diagnosis and location, a genetic algorithm based M-ary support vector machine (SVM) was proposed to enhance the accuracy of fault classification. The proposed method used the genetic algorithm and the within-class and between-class distance criterion to find the suboptimum fault coding which is related to the fault classes of power converters, and the optimized classifier model of M-ary SVM was established for the fault classification of power converters. The experiment results show that this optimized method has better fault classification accuracy than the standard M-ary SVM. Compared with the commonly used one-against-one SVM and one-against-all SVM, the proposed method needs far less number of classifiers and hence, the classification time can be reduced.