Research On Automatic Train Operation Of High Speed Train Based On Predictive Control Of Hybrid System Modeling

In recent years,high speed railways have been upgraded in the course of operation,and its technology has been continuously developed and improved.The high speed railway not only has special advantages in terms of passenger demand,but also has the advantages of large transportation capacity and high social benefits in terms of national strategy.High speed railway has become a common trend in the development of transportation in various countries.The Automatic Train Operation(ATO)system is an important subsystem in the Automatic Train Control(ATC)system.It is of great significance to coordinate the work between the systems,improve the potential efficiency of the transportation system,make the whole line of transportation safe,efficient and smooth operation,and realize the modernization of operation management.The research content of this thesis mainly includes the following aspects.Firstly,the Hybrid System(HS)theory is deeply analyzed,and the classification of the hybrid system modeling model is expounded and the relationship between them is revealed.The development status,theoretical basis and three main feature including prediction models,rolling optimization and feedback correction of Model Predictive Control(MPC)theory are systematically described.Secondly,the train dynamics equation is studied in depth,and the train dynamics model of hybrid system is established.Aiming at the nonlinear characteristics of running resistance,it is linearized by segmented linearization method,which not only simplifies the difficulty of controller design,but also preserves the nonlinear characteristics of running resistance.At the same time,integer variables are introduced to establish a segmented affine model of trains.In order to facilitate the solution of control law,the segmented affine model is transformed into mixed logic dynamic model,and the process of establishing dynamic model of train hybrid system and its mutual conversion process are described in detail.Thirdly,the performance index to meet the requirements of automatic driving is designed based on predictive control theory,including: punctuality,comfort and energy saving.In order to simplify the controller computation on the basis of ensuring the working performance of the controller,a stepwise control strategy is introduced,then the optimization problem is transformed into a mixed integer two programming problem to solve the control law,which effectively solves the constraint problem in the automatic driving control of the train.The effectiveness of the algorithm and the performance of the controller are verified by numerical simulation.Finally,aiming at the influence of complex and changeable environment and disturbance on train operation condition,in order to improve the tracking accuracy of train driving controller to target curve,a train prediction controller based on neural network feedback control is proposed,and the real-time error is studied and compensated online.This method was compared with the predictive control method which does not introduce feedback control,the simulation experiment is carried out,and the experimental results are analyzed.Simulation results show that this method has a significant effect on improving the tracking performance of train driving controller.