Gas Turbine Electric Actuator Fault Diagnosis Based On Deep Learning

Electric actuator is the key link to ensure the stable performance of gas turbine control system and reliable operation of equipment.As an indispensable component in the gas turbine control system,if the failure is not handled in time,it may lead to the failure of the whole system and even disastrous consequences.Therefore,the fault diagnosis of electric actuator of gas turbine control system has important engineering application value.This paper takes gas turbine electric actuator as the research object,focuses on typical fault signal processing method and deep neural network method,and adds Gabor filter to enhance feature extraction ability.Main achievements are as follows:(1)The conventional convolution neural network with the ability to process two-dimensional data is applied to one-dimensional data,in which the convolution and pooling operations are changed to one-dimensional convolution and one-dimensional pooling respectively.1dcnn is used to extract fault features and optimize parameters,which can effectively improve the efficiency of fault diagnosis;(2)Using the experimental data obtained from the actuator fault semi physical experimental platform,the nonlinear fault model of gas turbine electric actuator is established,and the relationship between residual signal and fault is established;The softmax classifier is trained.Finally,the one-dimensional convolutional neural network and softmax classifier are combined to diagnose the fault of gas turbine electric actuator.The simulation shows the effectiveness of this method;(3)A one-dimensional convolutional neural network(G-1DCNN)fault diagnosis model based on Gabor filter was established,which not only ensured the accuracy of fault diagnosis but also improved the operation speed of the model.The experiment shows that the G-1DCNN model can effectively improve the fault diagnosis of gas turbine electric actuator under the condition that the software and hardware conditions and training parameters remain unchanged.It has good diagnostic effect.