Regional traffic signal lamp control method based on graph neural network

Abstract

The invention provides a regional traffic signal lamp control method based on a graph neural network. A traffic flow predictor and a traffic signal lamp controller are trained at the same time, a future traffic flow change prediction value under a current intervention action is predicted by using the traffic flow predictor to help the traffic signal lamp controller to generate a new control scheme, the evaluation information of the traffic flow predictor to a value of the current action is used to assist in training the traffic signal lamp controller to maximize the long-term and short-term earnings of the traffic signal lamp control scheme. The traffic flow predictor and the signal lamp controller are built based on a depth message propagation graph network. According to the method, a system can be continuously optimized to adapt to changing traffic flows, and the road network smoothness degree and the traffic efficiency are improved.

Description

technical field [0001] The invention belongs to the field of traffic signal lamp control, in particular to a graph neural network-based regional traffic signal lamp control method. Background technique [0002] Traffic signal control is a critical yet challenging real-world problem, which aims to maximize the traffic efficiency of the road network and avoid possible traffic conflicts within intersections. In recent years, signalized intersections have become one of the biggest bottlenecks in improving traffic efficiency in urban traffic networks. Therefore, finding a feasible traffic signal control method that can automatically learn and adjust according to current and future traffic conditions can significantly alleviate traffic congestion and bring significant economic, environmental and social benefits. [0003] At present, traffic signal control systems are widely used in many modern cities, such as SCATS, SCOOT and other systems. The traffic signal schemes of these sys...

AI Technical Summary

Problems solved by technology

[0005] However, whether it is a traditional method or a method based on reinforcement learning, the main focus is on the independent control of one or a small number of traffic lights, ignoring the possible impact of traffic control behavior on the surrounding road network.

At the same time, the current method is still limited to the learning of historical traffic data and the timely response to the current traffic flow, ignoring the possible impact of traffic control behavior on future regional traffic flow

Images