一种基于支持向量机预测器模型的转子系统故障诊断方法  

Fault Diagnosis Approach for Rotor Systems Based on Support Vector Machine Predictive Model

作　　者：于德介[1] 陈淼峰[1] 程军圣[1] 杨宇[1] 

机构地区：[1]湖南大学,长沙410082

出　　处：《中国机械工程》2006年第7期696-699,共4页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(50275050);高等学校博士学科点专项科研基金资助项目(20020532024)

摘　　要：提出了一种基于支持向量机回归预测模型的转子系统故障诊断方法。分别对转子系统振动信号建立支持向量机回归预测模型,利用回归预测模型对振动测试信号进行预测,计算各支持向量机回归预测模型的预测信号与真实信号的误差并计算信噪比,通过比较各预测信号的信噪比来判断转子系统的工作状态和故障类型。实验结果表明,该方法能够有效地应用于转子系统的故障诊断。A fault diagnosis approach for rotor systems based on support vector machine predictive model was proposed. Firstly, the SVMs regression for predictive model of the vibration signals of a rotor system was established. Then, the support vector machine models were used to predict some time series of vibration signals and the values of SNR were calculated. By comparing the values of SNR, the conditions and fault patterns of a rotor system can be identified. Practical examples demon strate that the SVMs predictive model can be applied to rotor system fault diagnosis effectively.

关 键 词：回归 预测 支持向量机 故障珍断 转子系统 

分 类 号：TH113[机械工程—机械设计及理论]