High-speed train operation adjustment method and system based on Q learning

Abstract

The invention discloses a high-speed train operation adjustment method and system based on Q learning, and relates to the technical field of high-speed train dynamic scheduling. The state, action and strategy of the train operation adjustment process and a reward function with the minimum train delay time as an objective function are accurately described, and a delay train dynamic adjustment scheme under an emergency is obtained through an interactive learning mode to assist a dispatcher in making a decision. Furthermore in two aspects of time-space supplying and restriction of a high-speed railway network resource, A road network operation simulation module supporting machine-machine interaction is designed and built, normal operation and late operation scenes caused by typical emergencies can be simulated, a scheduling instruction automatically generated by a scheduling scheme can be rapidly received to simulate train operation, the dynamic change process of a high-speed railway network is described, and the feasibility of the scheduling scheme is verified. And finally, real operation scene data are input to obtain a scheduling scheme, the validity of the method and the system is verified, and a new solution idea is provided for the train dynamic operation adjustment method.

Description

technical field [0001] The invention relates to the technical field of high-speed rail dynamic dispatching, in particular to a Q-learning-based high-speed train operation adjustment method and system. Background technique [0002] Train operation adjustment is a core key link in railway transportation production. High-speed trains are easily affected by factors such as weather, equipment failure, and emergencies during operation, which will cause the actual running track of the train to deviate from the pre-established train operation plan. It is necessary to adjust the running time of the train in time to ensure that it can To maximize the fit with the train diagram, otherwise it will cause a series of unpredictable losses. A method is needed to obtain a better dispatching plan, and it will be simulated and verified through the existing dispatching system. At present, my country's high-speed railway mainly adopts three-level unified dispatching of "headquarters-railway bur...

AI Technical Summary

Problems solved by technology

The Q-learning algorithm is mainly used in intercity traffic coordination control problems and high-speed train energy-saving optimization problems. It is rarely used to solve train operation adjustment problems under emergencies. The solution efficiency becomes lower, and it is difficult to obtain a better solution to the problem

However, realizing the Q-learning algorithm requires train operation simulation software to have interactive capabilities. At present, there are many researches on train operation simulation software. Although they have very accurate ability to simulate train operation process, these simulation systems are not designed and developed for machine learning. , the simulation efficiency is slow, mainly human-computer interaction, lack of fast "machine-machine" interaction capabilities, not suitable for reinforcement learning that requires a lot of interaction and constantly changing operating scenarios

Images