Research On Fault Diagnosis Technology Of Electro-hydraulic Servo Valve Using Artificial Intelligence Algorithm

As a key component in the electro-hydraulic servo control system,the electro-hydraulic servo valve plays the role of "heart" in the entire servo system,and its performance affects the operation of the entire servo system.Since the electro-hydraulic servo valve is a highprecision component that integrates machinery,electricity,and hydraulics,its internal structure is relatively complicated and expensive,and its fault performance is usually nonlinear.Once a fault occurs,it will seriously affect the operation of the whole system will cause greater economic losses.Therefore,it is of great significance to construct an efficient and accurate intelligent diagnosis model to realize the intelligent fault diagnosis of the electro-hydraulic servo valve.The research work of this paper will focus on the intelligent fault diagnosis of electro-hydraulic servo valves.The research contents and innovations are as follows:(1)Design of servo valve characteristic detection system.First of all,according to the national standard of servo valve testing and the actual situation of the project,combined with the functional requirements of the test platform,the overall design framework of the test system is introduced;then,the detailed design of the calculation and selection of the components required by the test system and the structure of the test platform are completed.;Finally,the overall scheme of the measurement and control system is analyzed,the hydraulic system is controlled by PLC,and the software interface of the upper computer is written by using Lab View.(2)In view of the long training period of traditional algorithms,many adjustment parameters,small samples,and non-linear problems in the fault types and causes of electrohydraulic servo valves,a fault diagnosis model based on support vector machine(SVM)was proposed.The single model has the shortcomings of being easy to fall into local optimal solution and early convergence.In order to improve this shortcoming,the combined optimization of simulated annealing algorithm(SA)and particle swarm optimization algorithm(PSO)is used to solve the problem of SVM parameter optimization.(3)Aiming at the difficulty of feature selection in machine learning models,the convolutional neural network in deep learning is used to replace traditional manual feature selection.Considering the timing problem in fault prediction and how to effectively combine the front and rear information,a convolutional neural network(CNN)is proposed to be combined with a bidirectional recurrent neural network(BRNN),a bidirectional long shortterm memory(BLSTM),and a bidirectional gated recurrent unit(The combination of BGRU)endows the fault prediction model with sufficient extraction of time features to achieve accurate prediction of electro-hydraulic servo valve faults.(4)By comparing the CNN network and CNN＿BRNN/BLSTM/BGRU network in many aspects,it is found that the CNN＿BGRU network can effectively predict the failure of the electro-hydraulic servo valve.Aiming at the shortcomings of the CNN＿BGRU network,the attention mechanism(Attention)was introduced,and finally the prediction model of Cnn＿BGRU＿Attention was constructed.After experimental comparison,the overall prediction effect of the model is better than that of the CNN＿BGRU network.