基于对向传播神经网络的水电机组振动故障诊断研究

应用频谱法和对向传播神经网络分类器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,使构造的对向传播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。仿真结果表明,与常规方法相比,频谱分析与对向传播神经网络相结合的方法进行故障诊断简单有效,且具有良好的鲁棒性和泛化能力,是一种有效的诊断方法。

A Study of Vibration Fault Diagnosis of Hydro-Turbine Generating Unit Based on the CPN Network

The vibration fault diagnosis of hydro-turbine generating unit was investigated by the method of spectrum analysis and a counter propagation network(CPN) classifier.The feature of vibration signals in the frequency domain are attracted using the spectrum analysis,and then the spectrum feature vectors can be served as the learning samples.Via training,the CPN is able to reflect the mapping relationship between the spectrum feature vectors and the fault types whereby achieving the diagnosis purposes.The simulation results indicate that compared with the conventional method,the combination method of spectrum analysis and a CPN in fault diagnosis is effective and characterized by better robust and generalization capacity so that it is an effective diagnosis method.