A traffic signal adaptive control method based on deep reinforcement learning

Abstract

The present invention relates to the technical field of traffic control and artificial intelligence. A traffic signal adaptive control method based on deep reinforcement learning comprises the following steps: (1) defining a traffic signal control agent, a state space S, an action space A and a reward function r, (2) Pre-training of the deep neural network, (3) training of the neural network using a deep reinforcement learning method, and (4) traffic signal control based on the trained deep neural network. By preprocessing the traffic data collected by magnetic sensing, video, RFID, and Internet of Vehicles, etc., the low-level representation of the traffic state including vehicle location information is obtained; secondly, the traffic state is perceived by the multi-layer perceptron of deep learning, and the obtained The high-level abstract features of the current traffic state; on this basis, use the decision-making ability of reinforcement learning to select the appropriate timing scheme according to the high-level abstract features of the current traffic state, and realize the adaptive control of traffic signals to reduce vehicle travel time and ensure traffic safety. Smooth, orderly and efficient operation.

Description

technical field [0001] The invention relates to a traffic signal adaptive control method based on deep reinforcement learning, which belongs to the technical fields of traffic control and artificial intelligence. Background technique [0002] With the increase of car ownership, traffic congestion has become a problem that plagues my country's economic development. In 2016, the per capita economic loss caused by traffic congestion reached 8,000 yuan. At the same time, it also brings energy problems, traffic safety problems and environmental problems. There are two options for solving traffic congestion, increasing traffic infrastructure construction and developing advanced traffic control systems. However, urban land resources are tight and limited, and simply relying on increasing traffic infrastructure cannot solve the problem. Therefore, more attention should be paid to the development of advanced traffic control systems. Currently, mainstream traffic control systems inc...

AI Technical Summary

Problems solved by technology

On this basis, an effective traffic signal control strategy is formulated through reinforcement learning, which solves the shortcomings of the traditional traffic signal control system, such as incomplete perception of traffic conditions and inaccurate formulation of signal control strategies, and realizes adaptive control of traffic signals at intersections.

Images