针对齿轮振动信号故障特征微弱及故障样本不足,提出基于特征信息融合的小波-SVM(支持向量机)故障诊断方法,用于多类齿轮故障诊断.该方法采用离散小波变换对齿轮的振动信号进行处理来构造特征向量,将多路信号融合后输入到SVM的多故障分类器中进行故障识别.实验结果表明:该方法能够在训练样本数量少的情况下,快速获得良好的分类结果,且其故障诊断准确率在96.67％以上；峰值和峰值因子对齿轮故障最敏感,以峰值或峰值因子为特征量的多传感器信息融合,其诊断准确率达95％.该方法更适合于实际齿轮故障诊断应用,并为多类齿轮故障快速诊断的进一步创新研究提供了理论基础.%To solve the problems that the vibration signals from a gearbox are usually noisy and fault samples are usually insufficient, an intelligent diagnosis for gear fault identification based on feature-level information fusion by using discrete wavelet-SVM (support vector machine) is presented for multi-class gear fault diagnosis. According to this method, feature vectors were fused after extracted from gear vibration signals by discrete wavelet transform and they were input into a multiple-fault classifier of the support vector machine for fault identification. The experimental results show that this method can achieve precise classification quickly with small training samples and the diagnostic accuracy identification rate can reach 96. 67%. The peak and peak factor are the most sensitive characters for gear fault diagnosis and the diagnostic accuracy rate can reach 95% by using the characters to perform multi-sensor information fusion. This method is more suitable for practical application of gear fault diagnosis and can provide theoretical basis for the further innovation research on fast multi-class gear fault diagnosis.
展开▼
