Research On Fault Diagnosis Of The Electro-Hydraulic Actuators Based On PCA-SOM

With the continuous development of the aerospace industry,the research on the performance and reliability of EHA(Electro-Hydraulic Actuators),a key component in aircraft,has attracted much public attention.The research on EHA fault diagnosis method is helpful to understand the real-time running status of the mechanism,to give timely warning of faults,and thus to effectively improve the reliability of service equipment and to provide support for the following PHM(Prognostics and Health Management)technology research of EHA.Taking the EHA established by Beijing Academy of Precision Electromechanical Control Equipment as the research object,the main contents of this paper are as follows:1)On the basis of the EHA structure,this paper compares it with that of the traditional ones,and the ascending method is used to obtain all the basic events that affect the failure of the mechanism.Through the fault tree optimization and the reliability analysis of its structural function,the main failure modes and occurrence mechanism are listed.The unique weak link of EHA mechanism compared with other new servo mechanisms is found out,and finally the oil filter blockage is selected as the following fault research object.2)According to the working principle of EHA,the AMEsim simulation model of each component is established.Aimed at the problem of oil filter blockage fault setting,the mechanism and characteristics of the fault are analyzed,and the damping holes with different apertures are preset to simulate the oil filter blockage under different degrees and perform performance simulation.The simulation data show that the blockage of oil filter will lead to abnormal pressure in the hydraulic circuit,thus affecting the performance of EHA.3)Using the artificial neural network algorithm,based on the traditional SOM(Self Organizing Map)neural network algorithm,principal components analysis(PCA)is introduced,and the PCA-SOM oil filter blockage fault diagnosis method is proposed by setting the dimension reduction of the input data and the contribution rate of the neuron competition domain value.Compared the PCA-SOM neural network algorithm with that of traditional SOM,the efficiency and applicability of the new algorithm in terms of prediction accuracy and aggregation effect are verified.Further research can find that,for EHA,the traditional single measurement threshold method can only identify the faults of severe blockage states and its recognition accuracy is not high,while the PCA-SOM neural network algorithm has certain recognition ability for slight blockage states on the premise of improving the recognition accuracy.4)The prototype of the EHA is taken as an object,and various data signals such as displacement,speed,oil pressure and the like are collected.The PHM monitoring platform of the EHA is initially built through the simultaneous combination of the test platform,the back-end data storage and processing system,and the front-end interactive window application interface.