目前SVM核参数的选择是SVM实际应用的难点。通过分析SVM分类器原理,提出一种基于数据最大方差-熵准则的SVM核参数选择算法。该算法利用最大方差-熵准则衡量样本在特征空间中的线性可分性,然后结合PSO算法进行参数寻优,并将其用于双二次滤波电路的故障诊断中,优选得到的核参数提高了故障诊断的精度。 At present, the choice of SVM nuclear parameters is the practical difficulty of SVM. By analyzing the principle of SVM classifier, this paper proposes a SVM kernel parameter selection algorithm based on the maximum variance-entropy criterion of data. The algorithm uses the maximum variance-entropy criterion to measure the linear separability of the sample in the feature space, and then combines with the PSO algorithm to optimize the parameters and uses it in the fault diagnosis of the biquadratic filter circuit. The optimal kernel parameter is improved The accuracy of fault diagnosis.