Fault Diagnosis Of Valve-controlled Hydraulic Cylinder System Based On Neural Network

Aiming at the problem that the hydraulic components in the valve-controlled hydraulic cylinder system cannot be monitored and diagnosed in real time.The paper uses sensors to collect system signals and combines neural network methods to diagnose different types of leakage faults of hydraulic cylinders and early single-nozzle blockage faults of double-nozzle baffle servo valve.The main research of the paper is as follows:1.The failure mechanism of hydraulic cylinders and double-nozzle flapper servo valves is analyzed,and a hydraulic system failure simulation method is designed.The hydraulic cylinder structure design and external component method are used to realize the hydraulic cylinder internal and external leakage failure simulation of different degrees.Collect the pressure signal of the hydraulic cylinder and the displacement signal of the piston rod through the sensor,determine the representative fault characteristics and extract the corresponding characteristic value as the fault sample.After learning and training through the BP neural network learning algorithm,and using part of the sample as the detection data to verify Its feasibility.The results show that the BP neural network can accurately distinguish the internal and external leakage faults of the system and achieve the purpose of fault diagnosis.2.The hydraulic mechanical system modeling and simulation software AMESim is used to establish a dual-nozzle flapper servo valve-controlled symmetrical hydraulic cylinder position servo system simulation model.The displacement curve of the piston rod and the pressure of the left cavity of the hydraulic cylinder corresponding to the internal and external leakage of the hydraulic cylinder are obtained by the simulation.Comparing the curve with the corresponding curve obtained from the experiment,it is found that with the intensification of the failure degree,the curve change trend obtained by simulation and experiment is basically the same.Simultaneously,the single nozzle blockage fault of the double nozzle flapper servo valve is simulated,and the resulting servo valve has no input The state parameters in the signal state are consistent with the theoretical derivation,which verifies the reliability of the simulation model.3.Using the concept of representation learning to diagnose the early single-nozzle clogging fault of the double-nozzle flapper servo valve.Design a neural network to find a non-linear mapping relationship,through which the original data is transformed into another space that is easier to classify.By evaluating the contour coefficient of the data cluster formed by the data points,it shows that the clustering effect of the data points is good.The original data and new data are used to diagnose faults through BP neural network.The results show that the accuracy of fault diagnosis is greatly improved.Figure 52 table 9 reference 74...