基于SVDD的多设备故障源识别定位方法研究

Study on the Fault Sources Identification and Localization of Multi-Machine System Based on SVDD Method

对由多个设备组成的系统,检测的快速性和面临的小样本限制是进行声学故障源识别通常要考虑的两个重要问题。本文以线谱增强类声学故障为研究对象,提出了一种基于支持向量数据描述(SVDD)的故障源识别定位方法。该方法依据正常样本建立数据描述模板,对线谱增强类声学故障的出现进行识别,然后通过与各个设备上测得的传感器信号联合分析,实现故障设备的准确定位。实验结果表明该方法具有较好的工程应用性。

Rapidity of detection and limitation of small sample set are two important issues for identifying the acoustic fault sources of the system comprising multiple machines.In this paper,a new method based on Support Vector Data Description（SVDD）is presented to identify and localize the line-spectrum power-strengthening acoustic fault sources of the system.In this method,according to the normal-state samples,a data description template is established,and the line-spectrum power-strengthening acoustic fault sources are detected.Combining the fault detection result with the signal of sensors installed on each machine,the exact localization of the machinery fault is realized.The experiment results show that the method is feasible in engineering application.