基于多特征参数和概率神经网络的滚动轴承故障诊断方法

针对滚动轴承故障振动信号的非平稳特性，提出了一种基于多特征参数和概率神经网络的滚动轴承故障诊断方法。首先利用经验模态分解（EMD）方法将采集到的滚动轴承原始振动信号分解为有限个固有模式函数（IMF）之和，然后提取表征故障信息的若干个IMF的能量、峭度和偏度作为概率神经网络的输入参数来进行故障分类。试验结果表明，该方法可以准确、有效地识别滚动轴承的工作状态和故障类型，是一种可行的滚动轴承故障诊断方法。

Fault Diagnosis of Roller Bearings Based on Characteristic Parameters and Probabilistic Neural Network

According to the non-stationary characteristics of roller bearing fault vibration signals,a fault diagnosis approach based on on multiple characteristic parameters and probabilistic neural network was proposed. Firstly, original signals were decomposed into a finite number of stationary intrinsic mode functions(IMFs). Energy, kurtosis and skewness feature parameters were extracted from IMFs which contained main fault informations could be served as input parameters of neural networks to identify fault patterns of roller bearings. The experimental results show that the approach can identify working conditions and fault types of roller bearings.