基于多类SVM与改进EMD的故障诊断

Fault Diagnosis Based on Multi-class Support Vector Machine and Improved Empirical Mode Decomposition

鉴于传统方法在智能故障诊断中存在着一些不足,提出了一种基于多类支持向量机(SVM)和改进的经验模式分解(EMD)的故障检测与诊断办法。首先通过采用窗口平均法的EMD将原始信号自适应分解到分布在不同频带的基本模式分量(IMF),再用特征归一化处理进行特征提取,然后输入多类SVM分类器进行分类,从而对设备的当前状况作出判断。经过实验证明,本方法可以有效地对轴承设备进行故障诊断。

Aiming at the disadvantage of classic neural networks,a new fault diagnosis method is proposed based on multi-class support vector machine（SVM） and improved empirical mode decomposition（EMD）.Firstly,vibration signals are adaptively decomposed into several intrinsic mode functions（IMF） from original signals by EMD using window average.Then those functions,which belong to different frequency bands,are regarded as the input characteristic vectors of SVM for fault classification after dealing with the feature normalization Lastly,information is acquired for judging the status of devices.This method proved to be valid in a bearings fault diagnosis examples.