Research On Internal Leakage Detection Of Electro-hydrostatic Actuator Based On Operating State

The electro-hydrostatic actuator(EHA)is a new type of actuation system for aerospace vehicles,known for its high reliability and power-to-weight ratio.EHA typically operate under high speed,heavy loads,and inadequate cooling conditions.With increasing cumulative operating time,wear and aging of axial piston pump components and cylinder seals may lead to internal leakage in the hydraulic system,affecting performance and endangering the safety of the aerobat.Currently,there is neither direct observation nor effective indirect detection methods and means for EHA internal leakage.Difficulty in reducing the risk of failure.This article proposes an internal leakage detection method based on deep learning algorithm by studying the correlation between the operating status of the electro static pressure servo mechanism and the internal leakage of the electro-hydrostatic actuator.The main work and conclusions are as follows:(1)Overview of the development status of electro-hydrostatic actuator at home and abroad,analysis of the current state of research on the mechanism of internal leakage in hydraulic systems,an overview of the progress of domestic and international research on electro-hydrostatic actuator condition detection methods.(2)Using AMESim and Matlab/Simulink software platforms,a joint simulation model of a permanent magnet synchronous motor,a plunger pump,a hydraulic actuator and a hydraulic circuit was built.Through theoretical analysis and internal leakage simulation test,the variation law of motor current,rotational speed and hydraulic actuator inlet and outlet pressure with internal leakage is determined;a fault injection test technique is proposed to simulate internal leakage by installing a throttle valve,and the effectiveness of the method is verified by comparison between simulation models.(3)By adding a throttle valve between the inlet and outlet of the cylinder and the pump,an electro-hydrostatic actuator internal leakage fault simulation test system was built;the data of minor,moderate and severe internal leakage states were simulated and tested;the sliding average method was used to pre-process the sensing data for noise reduction,and the effectiveness of the internal leakage fault simulation technology was verified through the analysis of the test data.On this basis,a multi-source sensing signal sample library was constructed to provide data sources for the training and testing of the detection model.(4)A multiscale dense residual convolutional neural network is designed by introducing a residual network for the one-dimensional convolutional neural network that cannot comprehensively extract feature information from multi-source data;the parameters of the multiscale dense residual convolutional neural network model are determined by adaptive optimization of the hyperparameters of the neural network model using an improved sparrow search algorithm.By comparing with BP,SVM and WDCNN(Wide Convolutional Kernel Deep Convolutional Neural Network),the multi-scale dense residual convolutional neural network model can effectively improve the detection accuracy and generalization performance in solving the problem of internal leakage detection under variable work,and the detection accuracy of the test set reaches 95.2%.The proposed method provides a technical tool for internal leakage detection in EHA.