超临界350 MW机组汽轮机给水泵变频运行共振诊断方法

Frequency Conversion Resonance Diagnosis Method of Steam TurbineFeedwater Pump in Supercritical 350 MW Unit

研究超临界350 MW机组汽轮机给水泵变频运行共振诊断方法,解决给水泵变频运行时容易发生结构共振,影响机组运行可靠性的问题.选取最大振动读数位置,在超临界350 MW机组汽轮机给水泵电动机上部轴承位置布置振动传感器,采集水平方向与垂直方向的水泵振动信号.利用小波灰度矩方法从所采集振动信号中,提取汽轮机给水泵变频运行共振特征.将此共振特征作为径向基函数(RBF)神经网络的输入,RBF神经网络的隐含层利用径向基函数处理共振特征样本,输出给水泵变频运行共振诊断结果 .结果表明,该方法可以精准诊断不同工况下的超临界350 MW机组汽轮机给水泵变频运行共振故障,为机组可靠运行提供依据.

The resonance diagnosis method of the variable frequency operation of the turbine feedwater pump of the supercritical 350 MW unit is studied to solve the problem that the structure resonance is easy to occur during the variable frequency operation of the feedwater pump and affects the operation reliability of the unit.Select the maximum vibration reading position,and arrange vibration sensors at the upper bearing position of the turbine feed pump motor of the supercritical 350 MW unit to collect the horizontal and vertical vibration signals of the water pump.The wavelet gray moment method is used to extract the resonance characteristics of variable frequency operation of steam turbine feed pump from the collected vibration signals.The extracted resonance characteristics of the variable-frequency operation of the feedwater pump are used as the input of the RBF neural network.The hidden layer of the RBF neural network uses the radial basis function to process the resonance characteristic samples and output the resonance diagnosis results of the variable-frequency operation of the feed pump.The experimental results show that this method can accurately diagnose the resonance fault of the variable frequency operation of the turbine feed pump of the supercritical 350 MW unit under different working conditions,and provide a basis for the reliable operation of the unit.