A Joint Optimization Method Of Real-time Train Regulation And Passenger Flow Control For Fully Automatic Operation Systems

In recent years,with the rapid development of urban rail transit,it has become the most popular transportation in modern large cities.Train operation regulation is of great significance to ensure the safe and efficient operation of the metro.As one of the most critical factors of the operation regulation process,the number of people waiting on the station is coupled with the dynamic operation of the train.The rapidly increasing passenger flow makes the departure interval of the metro system smaller and smaller,which leads to the train operation susceptible to the external interference,thus causing the train delay and increasing the number of people waiting on the station.At the same time,the increase of the number of people waiting on the station will extend the dwell time of the train and strengthen the train delay.Therefore,it is essential to jointly optimize the train operation regulation with passenger flow control.With the widespread application of the fully automatic operation(FAO)system,the automation of the realtime train regulation is highly demanded.For the real-time train regulation optimization algorithms,most of the existing train operation regulation systems use centralized optimization algorithms,which have the disadvantages of low calculation efficiency and slow response speed and cannot meet the real-time requirements of train regulation.Based on the above description,this thesis carries out joint optimization research of real-time train regulation and passenger flow control,aimed at improving the on-time rate of train arrival and departure,train operation level,and service quality.This thesis first established a joint dynamic model of real-time train regulation,dynamic generation of train operation curves,and regulation of the number of people waiting on the station.Secondly,use distributed optimization algorithms to improve computational efficiency.Finally,use C# programming to design and implement the automatic train supervision(ATS)simulation verification platform,which mainly designs the operation diagram interface,the database,and the automatic train route.The thesis verifies the effectiveness of the distributed optimization algorithm for real-time train regulation and passenger flow control in the ATS simulation system.The research content and innovation of this thesis are mainly in the following aspects:(1)Considering the coupling relationship between dynamic train operation and passenger flow,establish a joint dynamic model of real-time train regulation,dynamic generation of train operation curves,and regulation of the number of people waiting on the station.On this basis,establish train dynamic regulation optimization objectives and operating constraints,and use the model predictive control(MPC)algorithm to transform the joint dynamic model into an optimized control model.(2)The thesis designs a distributed optimization algorithm of real-time train regulation and passenger flow control based on the alternating direction multiplier method(ADMM).ADMM algorithm can solve complex optimization models with coupling constraints.Compared with centralized optimization algorithms,this algorithm can effectively improve the calculation efficiency and ensure the real-time performance of train regulation.(3)This thesis verifies the distributed optimization algorithm of real-time train regulation and passenger flow control in the ATS simulation system and realize the function of automatically arrangement of train based on the improved A* algorithm and the real-time train regulation function based on the ADMM distributed algorithm in the ATS system.The effectiveness of the distributed algorithm of real-time train regulation and passenger flow control is verified through the effect of the actual train operation diagram being restored to the planned operation diagram after the train is disturbed.