Operation Curve Autonomous Planning And Distributed Cooperative Control Method For High-speed Trains

Nowadays,the construction of China’s high-speed railway has made remarkable achievements,and provided great convenience for citizens’ daily travel.As an important part of national economic development,high-speed railway plays an irreplaceable role in promoting the coordinated economic development of domestic underdeveloped areas.Although the domestic high-speed railway brings great convenience to people’s live,there are also some urgent problems to be solved.To meet the needs of economic development,the departure frequency of domestic high-speed railway is increasing significantly,multitrain,high-speed and high-density tracking operation in the same high-speed railway line has become an emerging normal,which is particularly prominent in some major railway networks,such as Beijing Shanghai high-speed railway.The existing train control methods are difficult to meet the safe,efficient,energy-saving and comfortable cooperative control of multiple trains at the same time,hence,the design of a train control strategy for multiple targets with operation curve autonomous planning function and cooperative control becomes a hot research issue in the field of automatic operation control of high-speed trains in the future.This study discusses the decision-making of high-speed train autonomous driving.Based on the operation characteristics of high-speed train and the coupling characteristics of multi-target tracking,the proposed methods mainly include the operation curve autonomous planning method based on deep learning(DL)and the distributed cooperative control strategy based on model predictive control(MPC)to improve the punctuality and flexibility of train operation within the overall safety of the railway,and the proposed distributed cooperative control scheme can also greatly reduce the computational burden of on-board computer.The main work of this paper is summarized as follows:(1)An advanced multi-train control framework based on deep learning is designed by combining distributed cooperative control method.As the problem that insufficient data in deep learning network training may lead to over fitting of prediction model,a data augmentation scheme based on generative adversarial networks(GANs)is proposed to generate data samples with the same distribution as the actual data samples to augment the training data set.(2)A hybrid reference trajectory learning model is proposed,which can scheme the train’s operation curve in real time from time-dependent and time-independent multiattribute data respectively.Subsequently,based on the model predictive control scheme,a distributed cooperative control optimization model is established considering the minimum tracking deviation of train operation,balancing safety,punctuality,energy efficiency and ride comfort.(3)To realize the decomposition of the optimization problem of high-speed trains,the dual decomposition technology is adopted to deal with the coupling constraints caused by the safety interval of adjacent trains,hence,each train can calculate the control strategy independently to reduce the workload of on-board computer and improve the robustness of the system.Finally,simulation experiments are conducted to verify the effectiveness and feasibility of the proposed method based on the actual application scenario.There are 38 pictures,11 tables and 62 references.