摘要
针对滚动轴承故障振动信号非平稳性、故障特征提取效果不理想以及故障诊断准确性低等问题,提出基于变分模态分解和多尺度排列熵的滚动轴承故障特征提取方法,并采用经粒子群算法优化的概率神经网络(PSO-PNN)故障诊断模型进行故障类型识别。通过变分模态分解方法将提取的振动信号分解成K个模态分量,进一步计算K个分量的多尺度排列熵,组成多尺度的特征向量,将特征向量输入到PSO-PNN故障诊断模型中识别故障类型。MATLAB仿真结果表明,该方法使故障类型识别准确率有所提高。
Aiming at the problems of non-stationarity of rolling bearing faulted vibration signal,unsatisfactory effect of fault feature extraction and low accuracy of fault diagnosis of rolling bearings,a method of fault feature extraction of rolling bearings based on variational mode decomposition and multi-scale permutation entropy is proposed.The probabilistic neural network(PNN)fault diagnosis model optimized by particle swarm optimization(PSO)is used to identify the fault types.The extracted vibration signal is decomposed into K modal components by using the variational mode decomposition method.The multi-scale permutation entropy of the K components is further calculated to form multi-scale feature vectors.The fea-ture vectors are input into the PSO-PNN fault diagnosis model to identify the fault types.The simulation results of MATLAB show that the method improves the accuracy of fault type identification of rolling bearings.
作者
张建财
高军伟
ZHANG Jiancai;GAO Junwei(College of Automation,Qingdao University,Qingdao 266071,Shandong China)
出处
《噪声与振动控制》
CSCD
2019年第6期181-186,共6页
Noise and Vibration Control
基金
山东省自然科学基金资助项目(ZR2019MF063)
山东省重点研发计划资助项目(2017GGX10115)
关键词
故障诊断
变分模态分解
多尺度排列熵
粒子群算法
概率神经网络
fault diagnosis
variational modal decomposition
multi-scale permutation entropy
particle swarm optimi zation
probabilistic neural network
