滚动轴承VMD能量熵与PNN故障模式识别研究

Research on VMD Energy Entropy and PNN Fault Pattern Recognition of Rolling Bearings

针对研究振动信号分析识别轴承状态的方法,在实践应用中受到各种噪声的影响很难达到准确识别预期目标的效果,提出了基于VMD能量熵特征与PNN神经网络结合的分类滚动轴承故障状态的方法。首先,通过运用变分模态分解(VMD)的信号预处理方法,实现振动信号的VMD降噪,同时利用集合经验模态分解(EEMD)对仿真信号进行对比两种方法的分解效果;然后,通过VMD能量熵和时域特征组成特征向量。最后,特征向量导入概率神经网络模型中准确识别滚动轴承故障状态。结果表明,该方法能将非平稳振动信号分解有效降噪且抑制模态混叠现象,同时能有效识别故障状态,对于在线监测机床健康状态领域的发展有重大的意义。

Aiming at the method of studying the vibration signal analysis and identifying the bearing state,it is difficult to achieve the effect of accurately identifying the expected target by the influence of various noises in practice.A method for classifying the fault state of rolling bearing based on the combination of VMD energy entropy feature and PNN neural network is proposed.Firstly,the VMD denoising of the vibration signal is realized by using the signal preprocessing method of variational mode decomposition(VMD),and the decomposition effect of the two methods is compared by using the empirical mode decomposition(EEMD);Then,Time-frequency domain feature vectors composed of VMD energy entropy and time-domain features.Finally,the feature vector is introduced into the probabilistic neural network model to accurately identify the rolling bearing fault state.The results show that the proposed method can decompose the non-stationary vibration signal effectively and reduce the modal aliasing phenomenon,and can effectively identify the fault state,which is of great significance for the development of on-line monitoring of the health state of machine tools.