Research On Control Strategy Of Electro-hydraulic Servo System Based On Data Fusion And Neural Network

Electro-hydraulic load simulator can simulate aircraft in special conditions,ships,cars and other large machines in different circumstances under load,the load simulator,the vast majority are electric hydraulic servo system and electro-hydraulic position servo system of load simulator,therefore to study is a matter of great significance.In which,the electro-hydraulic servo system is often rigidly connected with the position servo system,which will inevitably appear the position disturbance phenomenon in the process of simulating load.That is,in the loading process of the force servo system,the position system has a great influence on it due to its own movement,thus reducing the control precision of the force servo system.This academic dissertation focuses on the position disturbance of electro-hydraulic servo system,and studies how to restrain or eliminate the disturbance and improve the accuracy of the electro hydraulic servo system control so that improve the accuracy of the electro-hydraulic load simulator,decreasing the load error,this topic mainly comes from aerospace hospital project steering gear design of experimental platform.Mainly done and several aspects:(1)The mathematical model of electro-hydraulic servo system and position servo system is mainly established,and the mathematical model and transfer function diagram of the overall servo experimental platform are successfully established.On this basis,the noise of multiple sensors is measured and the influence of the noise on the system is considered,and the preliminary simulation of the system is carried out.The simulation results show that the tracking performance of the system is greatly improved by using multiple sensors.(2)The overall simulation model of the system is established in Simulink software,and the definition of excess force and the reason of excess force are explained.The control strategy based on the principle of structural invariance is used to suppress the superfluous force,and the amplitude and phase lag of the speed signal filtered by the filter will occur on the basis of the noise carried by the sensor,so that the control effect is not obvious after the application of structural invariance.(3)mainly introduces the multi-sensor information fusion and neural network method,and further optimize the BP algorithm learning BP-LM algorithm.Then,information fusion and neural network are applied to the steering gear experiment platform to obtain the speed signal without amplitude reduction and phase lag,and the position disturbance is suppressed by structural invariance.Finally,it is found that the BP neural network after LM algorithm has a better fitting effect on the speed,and the final application also verifies that the neural network method can obtain the ideal speed value more effectively than the filtering method,and the error between the neural network output force and the ideal output force is less than 0.5%.(4)In order to better verify the effectiveness of the neural network method,the above model is further simulated in the co-simulation of AMEsim and MATLAB.The simulation results show that this method is effective in simulation and can be further studied,and it can still achieve high precision force tracking with low error for complex input signals of simulated load station under specific conditions.