Fault Diagnosis of Gas Turbine Based on Support Vector Machine

In this paper, a fault diagnosis method based on support vector machine (SVM) is proposed for gas turbine bearing. Firstly, through analysis and processing of vibration signals, the singular value decomposition related EEMD technique is applied to extract feature vectors of the signals. The results are used as the input of SVM classifier model. Then, by using the SVM network intelligence, the turbine bearing operating status and fault type are determined. Experimental results show that the proposed SVM classification method with small sample can accurately and efficiently classify the working status and fault type of the gas turbine bearing, and has some engineering applications values.