基于KPCA-RVM的转子故障诊断

Rotor Fault Diagnosis Based on KPCA-RVM

针对转子故障振动信号特点,提出了一种基于核主成分分析及相关向量机(KPCA-RVM)的故障诊断方法。首先对故障信号用核主成分分析的方法进行降维处理以去除冗余信息以提高对数据进行计算处理的速度及正确率。之后使用相关向量机的方法对特征信息进行分类识别,以识别转子的正常、不对中、不平衡、碰磨以及松动五种不同运行状态。最后将本文所使用的方法与3种其他常见模型方法进行对比,结果表明本方法在转子故障识别上具有良好的可行性以及更好的实用性。

Base on the characteristics of the rotor fault, a method based on Kernel Principal Component Analysis and Relevant Vector Machine to diagnose rotor fault has been proposed. In this essay, firstly KPCA has been used to reduce the dimension of fault signals to remove redundancy information and improve the accuracy and computation speed, then RVM has been used to diagnose fault state which including normal condition, misalignment, unbalance, friction and looseness, finally, three other methods have been used to comprise with the method which is provided, and indicate the feasibility and practicability of the provided method in rotor fault diagnosis.