基于SVM的柴油机故障诊断方法研究

介绍了基于支持向量机(SVM)的故障诊断方法的原理和算法。利用小波包分解提取信号的特征参数,将特征送入故障分类器中训练。实验结果表明,当数据样本较少时,相比神经网络而言,基于SVM的故障分类器仍能正确分类多种故障。这种诊断方法具有算法简单、故障分类能力强的优点。

Research on Diagnosis Method of Diesel Fault Based on SVM

The principle and algorithm of the method of fault diagnosis based on support vector machine was introduced. The desired parameter of signal was extracted by wavelet packets' decomposition, and the desired parameter was trained by the fault classifier. The results show that compared to neural network, the fault classifier can still correctly classify multiple faults even when data sam-ple is limit. This method has simple algorithm and excellent capability of fault classification.