Traffic signal self-adaptation control system and method

Abstract

The invention discloses a traffic signal self-adaptation control system and method. For traffic control complexity, a fuzzy technology is adopted for designing a learning system reward and punishmentmechanism, and traffic state changes are more accurately and reasonably reflected; meanwhile, experience-based state division is utilized for constructing the Q-learning state space, the Q-learning state space updating complexity is lowered on the premise that multi-parameter traffic state evaluation is kept in the mode of building a traffic parameter fusion function, a phase-based green light timing plan is given, and therefore real-time response control over traffic flow is finally achieved. The traffic jam response time can be effectively shortened, signal control over various phases is rapid coordinated, and the intersection traffic efficiency is improved. Due to the model-free characteristics, the high self-adaptation capability and universality are achieved; the storage form of parameter indexes in a Q table is simplified in the system, the learning effect and response speed of the system for the traffic state are achieved at the same time, and the control complexity is lowered.

Description

technical field [0001] The invention belongs to the technical field of intelligent transportation, and in particular relates to a traffic signal self-adaptive control system and method. Background technique [0002] At present, urban traffic flow is growing at a high speed. With the rapid development of traffic, there is an urgent need for effective intelligent traffic management and control technology. In recent years, artificial intelligence technology has made great progress in the application of traffic control, providing new technical solutions for solving urban traffic control problems. Among them, the adaptive control system can adjust the timing parameters in real time according to the manager's control objectives and the characteristics of the time-varying traffic flow at the intersection. Compared with timing and induction control, it can make better use of the overall traffic capacity of the road network and effectively improve the traffic efficiency of the road n...

AI Technical Summary

Problems solved by technology

[0004] However, there are still some deficiencies in the application of Q-learning algorithm in traffic control

The existing Q-learning intersection traffic adaptive control method does not make full use of various traffic state parameters in complex traffic conditions, and it is difficult to give accurate and reasonable feedback for changes in traffic flow; and for the variable-period Q-learning adaptive control , its state space is too large, resulting in low learning efficiency and difficult to form effective control

Images