汽轮发电机组神经网络和模糊评判诊断方法研究

A Research on Fault Diagnosis for Turbo-Generator Based on Neural Network and Fuzzy Theory

将汽轮发电机组故障的实际因素和采集到的数据综合,并进行模糊化处理,反映了实际的运行过程,每个故障类型采集20组数据进行计算,其结果作为该故障类型的特征样本,并作为BP神经网络的输入进行模拟仿真,仿真结果与实际试验结果相符.克服了单一的根据频谱变化来判断故障类型的不足,为今后的工作提供了一种参考方法.

The actual factors of turbo-generator fault and the collected data are combined,and they are processed with fuzzy theory,which reflects the actual operation process.There are 20 sets of data collected to calculate for each fault.The result of each fault type will serve as the sample characteristic,and the input of the BP neural network for simulation.Simulation results are consistent with the real experiment results.It overcomes the shortages to estimate the fault type only by the variety of spectrum,and provides a reference method for the future work.