Iterative tuning strategy for setting phase splits with anticipation of traffic demand in urban traffic network

Facing large-scale urban traffic network, countless effort has been made toward intelligent and efficient urban traffic control system to better use existing traffic infrastructures. Recently, a novelty pre-timed traffic signal control strategy known as iterative tuning (IT) has been developed by exploiting repetitive characteristic of junction's vehicle throughput on working days, which is sufficiently efficient in under-saturated traffic conditions. This study further improves IT strategy in saturated traffic conditions with consideration of traffic demand including vehicle throughput and residual queued vehicles. Unlike conventional pre-timed strategies, preparation of signal schedules is not required in IT strategy and fine-tuning process executes iteratively and automatically. This study proposes a generalised traffic model to describe urban network dynamics and explicit split tuning algorithm. Without specific control trajectories, rigorous analysis generates sufficient condition for guaranteeing the convergence of IT strategy globally over repetitions. The robustness of IT controller to variations of traffic flow patterns and errors of initial conditions is also analysed in details. Commonwealth Avenue, an area with heavy traffic in Singapore, is demonstrated in simulations and simulation results indicate the effectiveness and robustness of IT strategy.