Data-driven Iterative Learning Queuing Length Control Based on Vehicle-road Coordination Systems

Abstract

Aiming at the intersection queuing length control, this paper establishes a model-free adaptive iterative learning control (MFAILC) scheme based on vehicle-road coordination systems. The MFAILC method is combined with the idea of vehicle speed guidance for the first time to deal with the traffic problem. Through the data communication between the vehicle and the road, a signal light control scheme is obtained, and then the vehicle guidance speed is given. The control objective is to make the queuing length at each intersection the shortest or the same. Firstly, the vehicle speed guidance scheme is given and the classical intersection model is analyzed. Since only controlling a single intersection does not play a great role in the traffic system, the research object of this paper is multi-intersection. Secondly, considering the complexity and repeatability of the intersection, its dynamic model is difficult to construct, so a MFAILC algorithm is presented. Then, the convergence of the proposed MFAILC scheme is derived. Finally, the feasibility of the MFAILC approach is verified by comparing the numerical simulation results.