基于小波变换的风力发电机组故障检测方法

Fault Detection Method of Wind Turbines Based on Wavelet Transform

针对传统风力发电机组故障检测方法受到非平稳振动信号影响,导致检测结果不精准的问题,提出了基于小波变换的风力发电机组故障检测方法。根据风力发电机组轴承非平稳信号特征,使用小波变换降噪技术,分解非平稳信号,获取有限长度离散含噪信号。消除噪声项后,利用峭度对非平稳信号的敏感性,提取故障自旋频率特征,实现轴承的故障检测。利用卷积神经网络提取齿轮箱阶次信号时序特征,通过齿轮箱故障时序特征的小波变换平移,利用阶次跟踪分析方法推导不同转速级的故障特征,以此对非平稳工况下齿轮箱故障状态诊断。由实验结果可知,该方法内、外滚道加速度时域信号变化范围分别为-0.3~0.3、-0.06~0.05 m/s 2,小、大齿轮断齿故障幅值为0.2、0.4,轴承故障和齿轮箱故障变化范围均与实际范围一致。

Aiming at the problem that the traditional fault detection methods of wind turbines are affected by the non-stationary vibration signal and result in inaccurate detection results,a fault detection method based on wavelet transform is proposed.According to the characteristics of non-stationary signals of wind turbine bearings,the wavelet transform noise reduction technology is used to decompose non-stationary signals and obtain discrete noise-containing signals of limited length.After eliminating the noise term,the sensitivity of the steepness to the non-stationary signal is used to extract the fault spin frequency characteristics to realize the fault detection of bearing.The convolutional neural network is used to extract the timing characteristics of gearbox order signal,and by the wavelet transform translation of gearbox fault timing characteristics,the order tracking analysis method is used to derive the fault characteristics of different speed levels,so as to diagnose the fault state of gearbox under non-stable working conditions.From the experimental results,it can be seen that the time domain signal variation ranges of internal and external raceway acceleration in the method are-0.3-0.3 m/s 2 and-0.06-0.05 m/s 2,respectively,the amplitude of small and large gear broken teeth is 0.2 and 0.4,and the ranges of bearing failure and gearbox failure are consistent with actual ranges.