基于频谱法与径向基函数网络的水电机组振动故障诊断

Vibrant Fault Diagnosis of Hydro-turbine Generating Unit Base on Spectrum Analysis and RBF Network Method

引起水电机组振动的原因很复杂,而且水电机组的振动故障往往是多故障同时发生,使得故障诊断很困难,目前主要是应用基于模式识别的神经网络来进行故障分类,尤其是BP网络应用较多。文章提出应用频谱法与径向基神经网络相结合的方法对水电机组的振动故障进行诊断。采用对水电机组振动信号进行频谱分析,提取该信号在频率域的特征量,将频谱特征向量作为学习样本,通过训练,使神经网络能够反映频谱特征向量和故障类型的映射关系,从而达到故障诊断的目的。水电机组振动故障诊断仿真分析表明,与常规方法相比,利用频谱分析和神经网络相结合的方法进行故障诊断具有简单有效等优点。

As the complexity of vibrant reasons and the multi-fault of coupling vibration, the vibrant fault diagnosis is very difficulty. The vibrant fault diagnosis of Hydro-turbine Generating Unit is distinguished by artificial neural network based on patterns recognition at present, especially by back propagation （BP） neural network. The method of spectrum and radial basis function （RBF） network was investigated for the vibration fault diagnosis of Hydro-turbine Generating Unit. Applying spectrum analysis to get the characteristics of this signal in frequency domain for the unit, and then using them as learning samples to train the network for realizing the mapping relationship between the fault and the spectrum characteristic, this method can be used for diagnosis of the unit faults efficiently. The diagnosis simulation experiments show that the proposed method can be diagnosed the faults of Hydro-turbine Generating Unit easily and efficiently.