ICA在汽轮机组动静碰磨故障诊断中的应用研究

摘要

针对旋转机械振动监测和故障诊断面临的噪声干扰和多信号混杂问题,将独立分量分析法(Independent Com-ponent Analysis,简称ICA)应用到汽轮发电机组振动信号分离上,该方法可将传感器所测的混合信号分离成相互独立的单个源信号,实现对故障源的准确识别,提高故障诊断精度.对多源信号混合-分离的仿真实验,成功验证了ICA法分离混合信号的有效性.采用ICA法对某台实际机组碰磨的轴振信号进行分离,结果从机组的碰磨信号中成功分离出了代表故障的周期性冲击信号,显示出ICA法对碰磨产生的冲击信号的分离效果,实现了对碰磨故障的诊断.%To solve the problem that the signals for vibration monitoring and fault diagnosis in field are always interfered by noises or other mechanical signals,a signal separating method called as Independent Component Analysis (ICA) is applied on fault diagnosis of turbo-generator units in this paper.The method can separated the mixed-signals into independent original signals,and it can identify the fault exactly and improve the accuracy of fault diagnosis.The simulation experiments for multi-source signals are studied,and the result shows the feasibility that the ICA can be applied to the signal separation for mixed-signals.The shaft vibration signals of rubbing from turbo-generator unit in field are processed by ICA.At last the periodic impulse signals are extracted out successfully from the rubbing signals of the unit.It shows that ICA is effective for the impulse signal to be separated from robbing signal and the fault is identified.