论文部分内容阅读
针对故障诊断系统中存在的大量无关或冗余的特征会严重影响故障诊断性能的缺陷,提出了基于交叉熵和支持向量机方法进行特征选择和参数优化的故障诊断方法.首先以某种概率分布产生若干随机样本,并依据交叉熵最小原理建立分布参数的更新规则进行特征搜索和SVM参数优化;然后利用优化后的特征向量和参数训练支持向量机获得故障诊断模型.故障诊断实验结果表明,该故障诊断方法能有效地优化故障特征和模型参数,提高故障诊断性能.
Aiming at the defect that a large number of irrelevant or redundant features in fault diagnosis system will seriously affect the performance of fault diagnosis, a fault diagnosis method based on cross-entropy and support vector machines is proposed to select features and optimize parameters.Firstly, with some probability distribution A number of random samples are generated and the update rule of distribution parameters is established according to the minimum entropy of entropy principle to search the features and optimize the SVM parameters.Then the fault diagnosis model is obtained by using the optimized eigenvector and parameter training SVM.The results of fault diagnosis experiment show that the Fault diagnosis method can effectively optimize fault characteristics and model parameters to improve fault diagnosis performance.