摘要
针对风电机组齿轮箱振动信号非平稳、不确定的特点,提出基于小波包与PCA遗传神经网络相结合的风电机组齿轮箱故障诊断方法。该方法选取振动信号的峭度和峰值作为时域特征值,利用小波包算法提取频带能量和二范数作为时频域特征值。考虑到特征值之间的相关性,利用主成分分析法确定主成分,从而减少神经网络的输入变量。利用遗传算法对BP神经网络权值和偏置进行优化,建立遗传神经网络的故障诊断模型。仿真测试结果证实了算法的有效性。
Aiming at vibration signal' s nonstationary and uncertainty of turbine gearbox, a fault diagnosis method based on the combination of wavelet packet and PCA genetic neural network was proposed,which takes vibration signal' s kurtosis and peak value as the time domain characteristic value, and makes use of wavelet packet algorithm to extract both band energy and 2-norm as the time-frequency domain characteristic value. Considering the correlation between the characteristic values,PCA (principal component analysis) was applied to determine principal component so that input variables of neural network can be reduced. By means of genetic algorithm, the weights and bias of BP neural network were optimized and the fault diagnosis model for genetic neural network was built. Simulation test proves effectiveness of this algorithm.
出处
《化工自动化及仪表》
CAS
2014年第2期144-148,194,共6页
Control and Instruments in Chemical Industry
基金
国家自然科学基金资助项目(61273144)
关键词
风电机组
齿轮箱
故障诊断
故障特征
小波包分解
遗传算法
BP神经网络
wind turbines, gearbox, fault diagnosis, fault feature, wavelet packet decomposition, genetic algo-rithm,BP neural network