An Iterative Learning Approach for Signal Control in Urban Traffic Networks

Traffic signal control influences route choice in traffic networks, and may even determine whether a traffic system settles in equilibrium or destabilizes into oscillatory patterns. Ideally, a stable equilibrium flow pattern should result from the interaction between control and route choice on a long-term horizon. This paper proposes an iterative learning approach for designing signal controls able to attract the system to equilibrium in an acceptable convergence speed. The traffic assignment model and combined traffic assignment and control problem are first introduced. An iterative learning control (ILC) based signal control is formulated and a basic model inversion method is analyzed. To deal with the nonlinearity of traffic system, a Newton based ILC algorithm is applied. Test in an example network verifies the effectiveness of the ILC method in achieving stable equilibrium in the traffic system.