基于遗传神经网络和证据理论融合的水电机组振动故障诊断研究

针对水电机组振动故障诊断中的故障误诊、漏诊以及诊断的可靠性低等问题,提出了适用于水电机组的神经网络局部诊断和证据理论融合决策诊断的故障诊断方法。在神经网络中应用遗传算法来提高网络的收敛速度,应用提出的诊断方法对水电机组振动故障进行仿真,诊断结果表明对故障征兆信息的有效组合,充分利用机组各部位的信息,可以减少诊断的误诊、漏诊问题,从而有效地提高诊断的可靠性。应用MATLAB7.0开发出故障诊断系统界面。

Fault diagnosis of the vibration of turbogenerator units based on GA neural network and evidence theory fusion

Aiming at fault misdiagnosis,missed diagnosis and low-reliability diagnosis existed in fault diagnosis of the vibration for turbogenerator units,a new fault diagnosis method acceptable for local diagnosis and evidence theory fusion diagnosis of turbogenerator units is highlighted.The circuital convergence rate is improved with GA method in neural network.Simulating the vibration fault of turbogenerator units with the method presented in the paper,the results show that effective combination of failure symptom information and full utilization of the message existed in every part of the units can decrease the misdiagnosis and missed diagnosis of problems so as to efficiently enhance the reliability of fault diagnosis.Therefore,MATLAB 7.0 is applied to develop the interface of fault diagnosis system.