基于信息熵和GA-SVM的自动机故障诊断

Automaton Fault Diagnosis Based on Information Entropy and GA-SVM

提出了一种运用信息熵和遗传算法优化的支持向量机(GA-SVM)对小口径火炮自动机进行故障诊断的方法。针对自动机工作时的短时冲击信号特征,首先运用具有自适应特性的局域波对信号进行分解得到IMF分量,并对各IMF分量进行Hilbert变换。接着利用信息熵理论提取局域波特征空间谱熵、边际谱熵和时频熵作为故障特征。最后将特征向量输入遗传算法优化的支持向量机进行故障分类识别。利用遗传算法的全局搜索能力对支持向量机的参数进行优化,摆脱了对求解模型的依赖。结果表明,相对于空间穷尽搜索寻找最优参数的支持向量机模型可提高诊断正确率。同时证明将信息熵和GA-SVM方法相结合在自动机故障诊断中的有效性。

A method which is used to diagnose the fault of small caliber artillery automation by applying the information entropy and the genetic algorithm optimization support vector machine （GA-SVM） is proposed.The shortterm impact signal feature is generated when the automaton works.Firstly,the local wave with self adaptive feature is used to break down the signal into a series of IMF components,and each IMF component undergoes the Hilbert transform.Then take advantage of the information entropy theory to extract the local wave feature space spectral entropy,the marginal spectrum entropy and time-frequency entropy as the fault features.Finally,the feature vector is input to the support vector machine with optimized genetic algorithm for fault classification and recognition.The use of global search ability of the genetic algorithm to optimize the parameters of support vector machine gets rid of dependence on solving model.The results show that this method relative to the space exhaustive search for finding the optimal parameters of support vector machine model can improve the diagnostic accuracy.Meanwhile,it proves that the combination of the information entropy and the GA-SVM method is valid in the automation fault diagnosis.