Subway departure scheduling method and system

Abstract

The invention discloses a subway departure scheduling method and system, and the method comprises the following steps: obtaining passenger riding information, line information and regular train information, and building a subway environment state matrix; obtaining metro reward and punishment data, and establishing a departure decision output by metro reward and punishment for judging a departure effect and generating a new departure decision; predicting a new environment state matrix at the next time point according to the subway environment state matrix and the riding information, and updating a departure decision output by the subway reward and punishment; inputting the subway environment state matrix and the reward and punishment data at each time point into a convolutional neural network for training, outputting a train departure decision after training, and executing subway departure according to the train departure decision. According to the method and the system, a Policy gradient algorithm is adopted to carry out reward and punishment output according to a subway environment state, intelligent optimization of subway scheduling can be realized, and an output result of output subway scheduling can more accord with a humanized design in combination with a convolutional neural network.

Description

technical field [0001] The invention relates to the field of subway control, in particular to a method and system for dispatching subway trains. Background technique [0002] The departure density of the subway in the stage is based on the real-time control of the service personnel in the subway station. When the passenger flow is dense, the passenger flow is first controlled by queuing, interception, etc., and the control of the departure is subjective and lagging. Moreover, the pressure on employment and cooperation is relatively high, and the experience of passengers during peak hours is poor. The premise of realizing train departure is passenger flow forecasting, and traditional passenger flow forecasting methods usually only have inbound passenger flow information prediction and outbound passenger flow prediction, and do not predict a complete itinerary of passengers, so the existing subway dispatching still exists and the overall situation Therefore, it is necessary t...

AI Technical Summary

Problems solved by technology

[0002] The departure density of the subway in the stage is based on the real-time control of the service personnel in the subway station. When the passenger flow is dense, the passenger flow is first controlled by queuing and interception. The control of the departure is subjective and lagging

In addition, the pressure on employment and cooperation is relatively high, and the experience of passengers during peak hours is poor.

The premise of realizing train departure is passenger flow forecasting, and traditional passenger flow forecasting methods usually only have inbound passenger flow information prediction and outbound passenger flow prediction, and do not predict a complete itinerary of passengers, so the existing subway dispatching still exists and the overall situation The passenger flow does not match the problem, so it is necessary to re-predict the flow of subway passengers in a special way

Images