关键词: 特征提取, 概率神经网络, 主轴承, 风力发电机, 故障诊断

Abstract: A fault diagnosis method based on feature extraction and an improved particle swarm optimization optimized probabilistic neural network is proposed to address the problems of nonlinear,non-stationary vibration signals and weak early fault characteristics of wind turbine main bearings under complex operating conditions. Wavelet packet denoising is used to preprocess the original vibration signals, and a total of 16 - dimensional feature vectors, including time-domain, frequencydomain,and operating parameters,are extracted to construct a multidimensional feature set that reflects the health status of the bearings. Secondly,an improved particle swarm optimization algorithm is adopted to adaptively optimize the smoothing factor of the probabilistic neural network, which improves classification accuracy and generalization ability. Experimental results based on historical vibration data of main bearings from ten power of 3. 0 MW wind turbines in a wind farm in Henan Province show that the proposed method achieves an average diagnostic accuracy of 94. 2% on the test set, which is significantly higher than those of the traditional probabilistic neural network,particle swarm optimization-back propagation neural network, and radial basis function neural network models. This verifies the effectiveness and superiority of the proposed method in wind power main bearing fault diagnosis.

Key words: feature extraction, probabilistic neural network, main bearing, wind turbine, fault diagnosis