On Several Train Trajectory Optimization Problems Based On Intelligent Algorithms

As the main artery of national economic development and comprehensive transporta?tion system,railway has the prominent characteristics of high transportation efficiency and low unit cost.Currently,the trains being operated on railway are operated by human train drivers under the supervision and protection of safety-critical equipments.However,with the expanding of railway network,the normalization of cross-line train running,the shortening of operative headway and the shifting of train operation speed,modern main line railway has proposed higher and higher requirements on the driving experience and skill level of train drivers.Meanwhile,the driving workload of train drivers are getting heavier and heavier,and the existing manual driving scheme gradually can not fully meet the practical needs.Researches on intelligent train operation methods and corresponding systems for human-machine cooperation have become a research hotspot in recent years and also an urgent problem to be solved in the field of practice.This thesis investigates the issue of train trajectory optimization based on intelligent algorithms.Key compo-nents such as model formulation,intelligent optimization algorithms,prototype function modules and experiment platform are studied from both theoretical and practical views.To be detailed,the innovations of this thesis are listed as follows:1.The train dynamics model and train trajectory optimization model are formulated.The major influence factors during the train operation process are firstly analysed,and a single-mass-point-based train dynamics model is formulated,taking the converted train length,train braking idling time and distance,self-use energy consumption into consid-eration.Based on the train dynamics model,a traction-distance-based train trajectory optimization model is built with the objective function of energy efficiency,punctuality and passenger comfort,meanwhile the corresponding constraint conditions and the opti-mality of trajectory are presented.2.An combinative algorithm with both off-line optimization and on-line optimiza-tion is proposed for inter-station train trajectory optimization.On the basis of the built mathematical models,the optimization algorithm is studied by using heuristic algorithm-s,numerical iteration algorithm and Pareto principle.Then case studies on only off-line optimization and combinative optimization of inter-station train trajectory are carried out and proved the adaptivity and efficiency of the proposed algorithms.3.An intelligent train trajectory optimization approach is proposed by combining machine learning and evolutionary computing methods.Deep neural network(DNN)and adaptive differential evolution algorithm(ADE)are used to calculate and to optimize the inter-station trajectory,respectively.Comparative experiments are then carried out considering different DNN types and structures with classic and adaptive DE algorithms.Results of different conditions are jointly analysed in aspects such as convergence pro-cess,optimal solution accuracy,etc.,to verify the effectiveness of the proposed approach.4.The intelligent driver advisory system considering real-time rescheduling infor-mation is proposed.Firstly the analysis of the existing information gap between current train scheduling and operation control is carried out and the framework of an ACP-based parallel intelligent system is built.Then key functional modules of iDAS are designed and developed on the basis of investigation on system architecture and function require-ments.Case studies on train trajectory optimization based on iDAS are then carried out to test the proposed approach.