A Fsat Generalized Predictive Control Algorithm For Controlling Braking Process Of High Speed EMU

The existing braking control of high-speed trains is manual.The control process requires high driving experience of EMU drivers.Often,the inappropriate operation of drivers causes the train to exceed the speed limit and the inaccurate parking position.It has important practical significance and application prospects to study the modeling and control method of braking process of high-speed train for automatic driving,and to realize automatic driving of train during braking process.Aiming at the structural and dynamic characteristics of the high-speed train's basic braking device,this paper establishes a multi-power unit model of the braking process of the EMU,which takes into account the time-varying parameters,time-delay and non-linear characteristics.Considering that the high-speed train consists of multiple braking units and the interaction of each braking unit,the multi-power unit model of the braking process of the EMU is established on the basis of the braking unit model Based on the multi-power unit model of braking system,a fast algorithm of MIMO generalized predictive control(GPC)based on Extreme Learning Machine(ELM)is used to track the braking speed of EMU.The braking process of EMU is punctual,comfortable,safe and accurate.The specific research work of this paper is as follows:1.Firstly,according to the dynamic characteristics of domestic high-speed trains,a distributed discrete model of multi-power unit is established considering the relationship between the coupling forces of EMU workshop.Then,according to the structure of the basic brake device of high-speed trains and the signal flow direction of the brake system,the brake unit of a single EMU is analyzed,considering the time-varying parameters,time-delay and non-linearity in the actual braking process of trains.The characteristic factors can be represented by a distributed autoregressive third-order discrete system,and the relationship between the input braking force and the output speed in the non-linear braking model of each power unit of EMU can be identified by the recursive least square method.2.Based on the multi-power unit model of high-speed EMU braking process,a fast algorithm of MIMO generalized predictive control based on limit learning machine(ELM)is designed to track the braking speed of EMU.Firstly,the input data(braking force)and output data(speed)are sampled at the current time.Then,the distributed autoregressive third-order discretization model and the model parameter vector identified by CARIMA are combined.Then,the model parameters are rapidly mapped to the generalized predictive controller coefficients by the limit learning machine(ELM),and the optimal control law is obtained.The increment of the control force at the sample time and the output tracking speed at the next time can be obtained.The fast tracking control of the braking speed of high-speed train can be realized by comparing the fast algorithm of EMU.The research results of this paper can provide a theoretical basis for the future high-speed EMUs to achieve safe and efficient ATO with high speed,punctuality and tracking accuracy.