Wind Turbine Generator Fault Detection Based on Multi-Layer Neural Network and Random Forest Algorithm

Abstract

In order to achieve fast and accurate fault diagnosis of wind turbine generators, a fault diagnosis algorithm based on multi-layer neural network (MNN) and random forest algorithm (RF) is designed based on the data of supervisory control and data acquisition (SCADA). The algorithm consists of two parts: The first part is the fault detection algorithm based on MNN. The reconstructed value of SCADA data is obtained by MNN, and the change trend of the reconstruction error and it exceeding the threshold are analyzed to predict the Wind Turbine (WT) fault and extract the fault sample. The second part is based on the random forest fault identification algorithm to establish an RF fault identification algorithm. The example verification proves that the MNN algorithm can detect the fault of the wind turbine generator earlier, and the RF identifies the fault type more accurately than other basic algorithms. It is concluded that the proposed diagnostic method based on multilayer neural network and random forest algorithm has good performance.