Fault Diagnosis of Wind Turbine Generator System Based on PMI-LSSVM

Abstract

Considering the multi-source wind power information such as wind speed, rotation speed, spindle horizontal and vertical vibration, a fault diagnosis method of wind turbine generator system based on partial mutual information (PMI) and least squares support vector machine (LSSVM) was proposed. A large amount of data containing fault status, such as blade fault, converter fault, generator fault, pitch bearing fault and yaw system fault, was analyzed. The PMI method was used to screen the characteristic parameters of the operation state of the wind turbine to identify the fault of the unit. The characteristic parameters of the wind turbine in various states were trained by LSSVM method to establish the mapping relationship between the parameter vectors of different characteristics and the fault types, so as to achieve the purpose of fault diagnosis. Besides, the different fault history data of wind turbine was used to test the fault model performance. The results compared with artificial neural network (ANN) method showed that the proposed method had good fault recognition ability and fast operation speed, which was suitable for fault diagnosis of multibrid technology wind turbine generator system, and can meet the requirements of online fault diagnosis.