Research On Predictive Control Method Of High Speed Train Based On Unmodeled Dynamics

At present,with the rapid development of all walks of life in our country,the transportation industry is also undergoing great changes with the momentum of prosperity.The construction of high-speed railway has also brought a better and better promoting effect to the economic development of the region and has been supported more and more widely.Under the premise of ensuring the safe operation of trains,how to improve the operating efficiency and performance more intelligently and conveniently has become the focus of attention.Therefore,the automatic driving control of trains has become an important research content in the field of high-speed trains.In the environment of rapid development of automatic train driving technology,this paper studies a speed tracking control method with better tracking performance.A high-speed train model is established to solve the problems of nonlinearity,time-varying parameters and coupling relationship between carriages in the process of high-speed train operation.Based on the model,the nonlinear generalized predictive control method and the nonlinear predictive decoupling control method are proposed.The specific research contents are as follows:1.The automatic driving technology of high-speed train includes the target planning of running speed and the tracking of the planned speed.Considering that many current control methods of speed tracking discuss the mathematical model which takes the control force as the input and ignores the dynamic process of the control force generation,this paper establishes the traction control model of the train from the perspective of the dynamic process of the control force generation/ In this paper,the influence of nonlinear factors on the train is described as unmodeled dynamics,and the train model is described as an integrated model composed of traction / braking model and unmodeled dynamics.2.Considering the unknown unmodeled dynamics of the system,this paper uses the neural network of extreme learning machine to estimate the unmodeled dynamics in real time.On the basis of the integrated model,the parameters of the model are estimated by the gradient identification algorithm,and a nonlinear generalized predictive controller with unmodeled compensator is designed to track the given target velocity.The stability and convergence of the proposed control algorithm are proved.3.Considering that the train is composed of several couplings,a multi-particle model of high-speed train is established by analyzing the force between the train cars and combining the relationship between the control force and the speed.Considering the coupling force between cars,the decoupling controller is designed on the basis of the nonlinear generalized predictive controller to eliminate the influence of coupling terms on the system.Taking CRH380 A high-speed train as the simulation object,the simulation experiment verifies the accuracy of the proposed method.