Research On Electro-hydraulic Load Simulator Based On Neural Network PID Control

The electro-hydraulic servo system has many good characteristics such as: high precision, fast responsibility, driving large inertia load, the ratio of power and mass is big and it's easy to combine with electric control, so it is widely used in the industry. The electro-hydraulic load system is an important part of the electro-hydraulic servo system, the electro-hydraulic load simulator studied in this thesis is an equipment that can simulate various aerodynamic load of the aircraft under laboratory conditions. Thus the classical self-destructing all-objective experiment will be converted to forecasting half-objective experiment in laboratory to achieve the aims such as shortening researching time, saving developing funds, enhancing reliability and success proportion.In the control fields, because neural networks is possessed of self-learning adaptive ability and mighty nonlinear match ability, it offers a effective solving method for nonlinear system modeling, but the single neural network control have the defects such as : low rate of convergence and it is easy to trap in load minimum problem. Conventional PID control based on accuracy model is possessed of simple structure, good robustness and high reliability, but it is difficulty to get the precise parameters. The conventional PID control is based on precise model, it's not easy to achieve good effectiveness using fixed parameters to adaptive system with variable parameters and many disturbance. Combining neural network technology with conventional PID control technology can utilize on-line identification of neural network to solve effective regulation problem of single-neural PID parameters, which can implement intelligence control.In this paper the author builds up mathematical model of electro-hydraulic load simulator and simplified it, and then takes deeply theoretical analyses on the control performances as well as their relations with structure parameters to provide the theoretical foundation of improving control performances of electro-hydraulic load simulator. A compound controller combined with the structure invariance principle and neural network PID control was adopted to enhance the system's loading ability. The structure invariance principle was adopted to do primary composition to surplus torque, then neural network PID was adopted to do secondary composition to surplus torque. Besides, this strategy can also be used to reduce disadvantageous influence caused by nonlinearities and uncertainties of electro-hydraulic servo system. Simulation and experiment show that the control strategy can improve dynamic characteristics of system, minimize stable state error of system, enhance adaptive ability and anti-disturbance ability of system, and effectively improve the load characteristics of electro-hydraulic load simulator.