An Optimal Control Method for Expressways Entering Ramps Metering Based on Q-Learning

In order to improve the using efficiency of expressways system, considering the features of the ramp control such as nonlinearity, complexity and uncertainty, a ramp controller based on Q-learning method is designed. In the case of the varying traffic demand, and with an aim at improving the traffic state performance of the expressways system, the optimum regulating rate of entering ramp is determined based on the Q-learning algorithm. An example shows that this algorithm can simply solve the problem of the optimum entering ramp metering in the expressways system in the case of traffic volume changes in demand.