基于第二代小波及PNN的自动机智能故障诊断  

Automaton Intelligent Fault Diagnosis Based on The Second Generation of Wavelet Transform and Probabilistic Neural Networks

作　　者：曹满亮 潘宏侠[1] 

机构地区：[1]中北大学机械与动力工程学院,太原030051

出　　处：《机械设计与研究》2015年第3期22-26,共5页Machine Design And Research

基　　金：国家自然科学基金资助项目(51175480)

摘　　要：提出了一种运用第二代小波信息熵和概率神经网络(PNN)相结合对自动机进行故障诊断的方法。针对自动机运动循环过程和振动信号的非线性短时冲击特性,提出了截取振动信号中有用信息进行第二代小波分析,在此基础上提取第二代小波熵,把其作为特征向量分别应用BP神经网络和概率神经网络对自动机进行故障模式识别。结果表明:概率神经网络相对于BP神经网络可以提高故障分类的正确率,从而验证了基于第二代小波信息熵及PNN在自动机故障诊断中的有效性。A method which is used to diagnose the fault of automation by applying the second generation wavelet information entropy and the probabilistic neural networks （PNN） is proposed. Combining automaton circle motion process analysis with the nonlinear, short-term impact characteristics of the vibration signal, this paper proposes the way of using the second generation of wavelet to analysis the intercepted useful information of the vibration signal. Based on this, the second generation of wavelet entropy parameters are extracted and then are regarded as feature vectors respectively applied to BP neural network and probabilistic neural networks for fault diagnosis of the automaton. The results show that PNN relative to BP neural network can improve the fault diagnosis accuracy. So it proves that the validity of the method of information entropy based on the second generation of wavelet transform and PNN for automaton fault diagnosis.

关 键 词：自动机 第二代小波变换 第二代小波熵 概率神经网络 BP神经网络 故障诊断 

分 类 号：TH17[机械工程—机械制造及自动化]