On the resilience-enhanced control for road traffic systems - An “ecological” approach

AbstractThe resilience of road systems is one key focus in transportation science literature. In particular, many studies have focused on traffic control development for improving the resilience where the mainstream control methodology is to drive the traffic system to an assumed single optimal equilibrium and a local recovery from failure or disruption. This setting appears to be oversimplified, as road traffic system can feature multiple (locally) stable states like many other complex physical systems, and thus there is lack of understanding on can and how a system transition between stable states. In Holling (1996) and Folke (2006), etc., such ability is defined as “ecological resilience” and it has not been applied within the context of urban road traffic control. In this work, we define urban traffic ecological resilience as the ability of a traffic system to absorb unpredictable congestion disturbances, by shifting to alternative steady states without the need of returning to the original non-congested equilibrium. A traffic flow control framework is developed to enable this ability, and it comprises three aspects: portraying the recoverable scopes via an “attraction region” modeling, designing alternative steady states around the classic equilibrium, and controlling towards local steady states while maintaining stable dynamics. Notably, this paper derives analytically the inner and outer of attraction region with explicit algebraic expressions, innovatively addressing the inherent complexity in attraction region derivation for nonlinear systems. Comparing to the classical resilience-oriented approaches, the proposed ecological resilience approach offers superior adaptability towards hyper-congestion; and enhances the overall system resilience, i.e., the recovery capacity from hyper-congestion. The proposed resilience-targeted control shifts from an engineering-focused paradigm to an “ecological” one, as the objective is not to move towards a rather infeasible optimality, but to facilitate an ecological transition among approachable local optimalities.