基于小波包和BP神经网络的风机齿轮箱故障诊断

摘要

为了对风力发电机组中最容易发生故障的核心部件齿轮箱进行故障诊断，提出基于小波包变换和BP （Back Propagation）神经网络的齿轮箱故障诊断方法。首先，根据齿轮箱工作时的振动信号特性，通过小波包变换方法对振动信号进行去噪、分解与重构，有效提取不同故障下各频段能量的故障特征；其次，将提取的能量故障特征输入至BP神经网络诊断系统中进行识别，实现故障的智能诊断。通过试验证明了该方法的有效性。%Gearbox is the core component of wind turbine, but it can be faulted easily. In order to monitor the gearbox, a fault diagnosis method based on wavelet packet transform and Back Propagation (BP) neural network was put forward. Firstly, the vibration signals of the gearbox were denoised, decomposed and reconstructed according to their characteristics using wavelet packet transform. Then, the fault features of the different frequency band energy were effectively extracted. Fi-nally, the fault energy features extracted were put into BP neural network diagnosis system to recognize the fault types. The system can implement intelligent fault diagnosis. The experiment demonstrated the efficiency of this method.