Design of networked freeway traffic controllers based on event‐triggered control concepts

Abstract

In this paper, an event-triggered model predictive networked scheme is proposed to control traffic in freeway networks. The freeway network is a spatially distributed system in which sensors and actuators act locally on portions of the overall system; measurements and control actions are transmitted to/from the feedback controllers via a shared communication network. In this sense, the considered plant can be properly controlled by relying on a design framework typical of networked control systems, a major feature of which is the necessity of taking care of both the computational effort and the communication effort. The proposed control scheme allows to reduce the computational load with a feedback model predictive controller in which suitable triggering conditions are defined. As for the communication effort, local triggering conditions are included in sensors, inducing transmissions of the system state only when such conditions are verified. The robustness properties of the controlled system are investigated in the paper by analyzing the input-to-state practical stability of the plant under the action of the proposed control strategy. The effectiveness of this proposal is also analyzed via simulation. 1163 efficient numerical solution algorithm has been adopted in the optimal freeway traffic control tool AMOC [5,6], by using the so-called feasible direction algorithm. An alternative approach is the receding-horizon based model predictive control (MPC), which, in the specific application case, needs to be applied in real time. In the literature, MPC has been extensively adopted to control freeway traffic. For instance, in [7,8], the nonlinear MPC framework is designed to include, as a predictor, the macroscopic model called METANET [9]. In other papers, for example [10,11], the considered prediction model is instead the cell transmission model (CTM) [12,13], suitably reformulated in mixed logical dynamical (MLD) form [14].A major feature of freeway traffic control systems that has not yet received significant attention is the necessity of considering not only the controller of the plant but also all the measurement equipment (sensors) and the way in which all the system components are connected and communicate. When a plant is a spatially distributed system in which the communication between feedback controllers, sensors, and actuators occurs via a shared communication network, the control system is classified as a networked control system [15,16].In this paper, we propose a control framework aimed at reducing congestion in freeway systems considered as networked control systems, referring in particular to the cases in which the communication between the sensors and the controller occurs through a shared digital communication network. In the design of the control scheme, specific attention is devoted to reduce the communication effort. More specifically, the number of state transmissions between the sensors and the controller is decreased by defining specific sensor triggering conditions in order to evaluate whe...

Event‐triggered MPC design for distributed systems toward global performance

“…In practice, many large‐scale systems are composed of dynamically coupled subsystems, such as process control systems, power grid systems, and the networked freeway traffic . In controlling such a large‐scale system, distributed model predictive control (DMPC) has attracted great attention especially since model predictive control (MPC) can deal with multivariate control problems and accommodate constraints explicitly.…”

“…In practice, many large-scale systems are composed of dynamically coupled subsystems, such as process control systems, 1 power grid systems, 2 and the networked freeway traffic. 3 In controlling such a large-scale system, distributed model predictive control (DMPC) has attracted great attention especially since model predictive control (MPC) can deal with multivariate control problems and accommodate constraints explicitly. In DMPC, each controller solves its own optimization problem based on the local measurement and information from a subset of other subsystems and communicates with these subsystems, by which the size of the overall optimization problem is reduced and a satisfactory global performance is achieved.…”

Adaptive model‐based event‐triggered sliding mode control

“…A similar approach is adopted in model-based networked control systems, which use an explicit model of the plant, asynchronously updated with the actual plant state when it is transmitted over the network [20]. The combination of the two previously mentioned approaches gives rise to the so-called model-based event-triggered control [21][22][23][24], which has satisfactorily resulted in being applicable to various fields (see, for instance [25][26][27]). …”

Editorial for the Special Issue on recent trends in traffic modelling and control

“…It is also worth noting that the modern vision of traffic control encompasses a strong integration between freeway traffic control systems and urban traffic management systems, in order to design efficient ‘smart mobility’ systems. For this reason, this Special Issue includes contributions relevant to both freeway traffic systems and urban traffic networks . These contributions are able to provide a significant scientific, technological, social and economic impact, with immediate effects on quality of life and environmental health, especially in terms of congestion reduction, efficient use of energy resources, reduction of costs associated with fuel consumption and wear of vehicles.…”

“…The first four contributions of the Special Issue are related to freeway traffic systems. Among them, Ferrara et al , propose an event‐triggered model predictive control scheme for a freeway network in which measurements and control actions are transmitted via a shared communication network. The proposed control scheme is designed in order to reduce the computational load of the controller and the communication effort between the sensors and the controller, by means of suitably defined triggering conditions.…”