基于核极限学习机的旋转机械故障诊断方法

摘要

Traditional fault diagnosis method of rotating machinery needs human intervention and has the problem of low diagnosis accuracy and unsteadiness.In order to diagnose the fault of the rotating machinery precisely and effectively,a fault diagnosis method of rotating machinery based on kernel extreme learning machine(KELM) is proposed. First of all, the vibration signals in the states of normal operation, rotor unbalance, rotor misalignment and bearing pedestal looseness are gathered.The wavelet energy of each frequency band is extracted as the fault feature from the denoised vibration signals of the rotating machinery. On this basis, the fault diagnosis model of the rotating machinery is established based on the KELM.The experiment on a rotating machinery fault simulation platform show that the KELM method has higher stability than ELM and the accuracy of KELM is above 99 %,which is better than support vector machine(SVM)and ELM.%传统的旋转机械故障诊断存在需人工干预、诊断精度低等问题.为更好诊断旋转机械故障,提出基于核极限学习机的旋转机械故障诊断方法.首先在正常状态和转子不平衡、转子不对中、轴承座松动3种故障状态下采集旋转机械的振动信号,降噪后提取各频段小波能量作为故障特征,以此为基础建立基于核极限学习机的旋转机械故障诊断模型.在旋转机械故障模拟实验台上进行的应用实验表明,KELM方法比ELM有更高的稳定性,KELM的故障诊断准确率大于99 %,诊断性能优于极限学习机和支持向量机.