A Deterministic and Stochastic Petri Net Model for Traffic-Responsive Signaling Control in Urban Areas

Abstract

The problem of reducing congestion within urban areas by means of a traffic-responsive control strategy is addressed in this paper. The model of an urban traffic network is microscopically represented by means of deterministic and stochastic Petri nets, which allow a compact representation of the dynamic traffic network. To properly model traffic congestion, intersections are divided into crossing sections, and roads have limited capacity. Each intersection includes a multiphase traffic signal, whose sequence of phases is given and represented by a timed Petri net. The control strategy proposed in this paper aims at minimizing queue lengths by optimizing the duration of each signal phase. This is accomplished by heuristically solving a stochastic optimization problem within a receding-horizon scheme, to take into account the actual traffic flow entering the network, thus making the proposed approach traffic-responsive. In this framework, the Petri nets play a key role, as the cost function to be minimized is a function of the marking, and the constraints include the marking state evolution. The proposed strategy is applicable to both undersaturated and oversaturated traffic conditions

Fuzzy Logic and Webster’s Optimal Cycle Based Decentralized Coordinated Adaptive Traffic Control Method

“…There are further researches performed in the field of traffic signal control using different methods and different optimization algorithms, such as Petri net [16], Bee colony optimization algorithm [17], deep learning algorithm [18], and Dynamic Programming [19]. The most of abovementioned methods either are non-cyclic strategies or applied to isolated intersections.…”

A Switching-Based Adaptive Dynamic Programming Method to Optimal Traffic Signaling

“…The first family is the microscopic simulation-based approach, which uses historical traffic data to build a vehicle-based simulation environment of the traffic network. Then, in combination with artificial intelligence learning methods, one can forecast the future states and design optimal traffic signal policies [10], [11], [12], [13], [14], [15], [16]. For example, researchers have proposed to control traffic lights in real time by means of reinforcement learning [12], [13], [14].…”

Time‐of‐day breakpoints optimisation through recursive time series partitioning

“…Most systems are built upon adaptive signal optimisation algorithm that optimises the signal plan in real‐time by using real‐time traffic data collected from the loop detectors. It is called a traffic responsive strategy [4–6]. However, in real life, the loop detectors often suffer from a high‐failure rate.…”