Vibration Fault Diagnosis of Hydroelectric Unit Based on LS-SVM and Information Fusion Technology

摘要

Vibration fault diagnosis of hydroelectric unit was investigated using method of least square support vector machine (LS-SVM) and Dempster-Shafer theory (D-S Theory). Spectrum and amplitude characteristic was acted as eigenvector of learning samples to train the constructed LS-SVM regression and classifier for realizing mapping relationship between the fault and the characteristic. Information fusion was realized after completing local diagnosis, and then fault diagnosis was achieved. Experiments show that the method has a rapidly diagnostic process and generalization performances. It is suitable for the vibration fault diagnosis of hydroelectric unit.