Research On Control System Of Tractor Hydrostatic Transmission

Hydrostatic transmissison(HST)is a completely enclosed hydraulic system,which mainly includes hydraulic pumps,motors and hydraulic control valves and other hydraulic components.Compared with traditional pure mechanical transmission and hydraulic transmission,HST has the advantages of simple structure,reliable work,and good speed regulation performance.It is widely used in agricultural machinery and engineering vehicles running at low and medium speeds.HST not only has good characteristics of stepless speed change and torque change,but also can carry out various speed regulation and control.The different control methods will directly affect the performance of HST.How to make HST regulation more accurate and rapid,effectively exert the working performance of HST,and improve the reliability of HST is one of the key issues in the research of hydrostatic transmission technology.The existing control methods of HST are mainly traditional control methods such as electro-hydraulic servo control,PID control and fuzzy control,and their control performance basically meets the requirements of HST normal operation.However,traditional control methods have poor performance in dynamic adjustment and multivariable control of HST,and cannot make the system have good static and dynamic characteristics and control flexibility.It is urgent to propose more effective control methods to improve the overall performance of HST.Aiming at this difficulty,back propagation neural network control(BP neural network control)and model predictive control(MPC)are used to study and optimize the control system of HST in tractor.Performance optimization.First of all,based on the mathematical model and transfer function of HST,the simulation model of HST is established in Matlab environment;three methods of PID control,fuzzy control and BP neural network control are compared and studied for the simulation test of the variable pump-quantitative motor system in HST.Obtain the advantages of BP neural network control in nonlinear system control,and then use BP neural network to control the variable pump-variable motor system in sections.Secondly,model the variable pump-quantitative motor system in the HST based on the state space equation,and use the particle swarm optimization algorithm to optimize the parameters of the uncertain hyperparameters(prediction time domain and control time domain)in MPC,in order to improve the performance of HST under the control of MPC.Finally,based on the built hydrostatic transmission test bench,the performance of HST under the control of MPC is verified by experiments.research shows:1)Compared with PID control and fuzzy control,BP neural network control can effectively improve the speed response of HST,reduce overshoot,and has better robustness.Sectional control of HST by BP neural network controller can increase the speed range of the system,reduce the speed fluctuation caused by the switching of the control phase and the load change,and increase the stability of the system.It has good guidance for the simultaneous control of VPVM system in the future.2)The HST controlled by MPC can complete the speed regulation process from zero to the expected value of 1200r/min in 0.72 s,and at the same time,the overshoot of the motor output speed is reduced to 1.25%,and the control performance of the system is significantly improved.For the expected speed of fixed length change in the range of 0-600r/min and different external loads,MPC has a faster response speed and a small overshoot,so that the system has good static and dynamic characteristics.3)In the bench test,the motor speed of the HST controlled by MPC can reach the desired speed of 500r/min from zero within 2.11 s,with an overshoot of 1.6%;the time for the motor speed of the HST controlled by BP neural network to stabilize is 2.23 s,and the overshoot of the system is almost zero.The experimental results of the two control methods are basically the same as the simulation test results,and both can better improve the responsiveness and stability of the system.At the same time,in the acceleration,deceleration or starting state,BP neural network and MPC have good control performance for HST,so that the speed is output smoothly with small fluctuations,achieving fast,accurate and small dynamic errors.