Local Model Predictive Control Research Articles

Distributed model predictive control over network information exchange for large-scale systems

A class of large scale systems, which is naturally divided into many smaller interacting subsystems, are usually controlled by a distributed or decentralized control framework. In this paper, a novel distributed model predictive control (MPC) is proposed for improving the performance of entire system. In which each subsystem is controlled by a local MPC and these controllers exchange a reduced set of information with each other by network. The optimization index of each local MPC considers not only the performance of the corresponding subsystem but also that of its neighbours. The proposed architecture guarantees satisfactory performance under strong interactions among subsystems. A stability analysis is presented for the unconstrained distributed MPC and the provided stability results can be employed for tuning the controller. Experiment of the application to accelerated cooling process in a test rig is provided for validating the efficiency of the proposed method.

Local model predictive controller in a solar desalination plant collector field

This paper proposes a new predictive control strategy for a distributed collector field of a solar desalination plant. The main purpose of the controller is to manipulate the water flow rate to maintain constant the outlet-inlet temperature gradient in the collectors in spite of disturbances. The controller is based on a filtered Smith predictor generalized predictive control algorithm and a simple procedure to update the linear model used in the predictor as well as the tuning parameters, in such a way that non-linear optimization is avoided. The controller copes with the process non-linearities, constraints, dead time and plant-model mismatch obtaining the desirable performance both, in the reference tracking and in the rejection of strong irradiance disturbances. Simulations and real experimental tests in AQUASOL desalination plant solar field are presented to show the advantages of the proposed controller.

Robust Distributed Predictive Control of Communicating and Constrained Systems

AbstractThis paper proposes a strategy for distributed model predictive control for linear time invariant systems coupled via a common cost function and constraints. Each subsystem solves a local robust model predictive control (RMPC) problem which considers uncertain predictions received from other subsystems. In the chosen communication scheme, subsystems exchange information about optimized local trajectories and time-varying disturbances sets obtained from the local RMPC. The latter optimizes over feedback policies based on a cost function motivated by min-max approaches and can be implemented as a single tractable quadratic program. Under some assumptions, the proposed scheme remains feasible even in the presence of delays and packet loss. Simulation results for a platoon of autonomous vehicles in a leader-follower scenario illustrate the approach.

Modelling and Predictive control of a Nonlinear System Using Local Model Network

AbstractIn this work the optimization of the local model network structure and predictive control that utilize the local model network to predict the future response of a plant is studied. The main idea is based on development of the local linear models for the whole operating range of the controlled process. The local models are identified from measured data using clustering and local least squares method. The nonlinear plant is then approximated by a set of locally valid sub-models, which are smoothly connected using the validity function. The manipulated variable adjustments are computed through optimization at each sampling interval. The parameters of the plant at each sampling point are derived from the linearization of local model network. The proposed identification and control method is illustrated by the simulation study on the nonlinear process.

Steering Vehicle Control by Switched Model Predictive Control

We propose a switching Model Predictive Control (MPC) strategy to control vehicle steering by actuating active front steering (AFS) and electronic stability control (ESC). After describing the piecewise affine prediction model used for MPC design, where the nonlinearities arise from the relation between sideslip angles and tire forces, a switching MPC strategy is implemented, where different local MPC controllers are used depending on the current tire force conditions. The designed controller maintains most of the benefits of a previously designed hybrid model predictive controller, but it has lower complexity and allows more flexible design. The controller stability is verified and the controller behavior during challenging step steering maneuvers is tested in closed-loop simulations against a nonlinear vehicle model.

Distributed model predictive control for plant-wide hot-rolled strip laminar cooling process

Since hot-rolled strip laminar cooling (HSLC) process is a large-scale, nonlinear system, a distributed model predictive control (DMPC) framework is proposed for computational reason and enhancing the precision and flexibility of control system. The overall system is divided into several interconnected subsystems and each subsystem is controlled by local model predictive control (MPC). These local MPCs cooperate with its neighbours through the scheme of neighbourhood optimization for the improvement of global performance. The state space representation of each subsystem’s prediction model is designed by finite volume method firstly, and then is linearized around the current operating point at each step to overcome the computational obstacle of nonlinear model. Moreover, since the strip temperature is measurable only at a few positions in water cooling section due to the difficult ambient conditions, an Extended Kalman Filter (EKF) is used to estimate the transient temperature of strip. Both simulation and experiment results prove the efficiency of the proposed method.

Experimental Local Model Predictive Control: Trajectory Tracking and Cost Function Tuning for WMR Reactive Navigation

This paper presents local model predictive control techniques for WMR (wheeled mobile robot) reactive navigation. The use of dynamic models and experimental cost function factor adjustments are important issues of the above work. In the methodology presented dynamic experimental models are used. In order to do this, a set of dynamic models obtained from experimental robot system identification are used to predict the horizon of available coordinates. Local trajectory planning is a relevant idea of this work. Hence, when dynamic environments or obstacle avoidance policies are considered, the navigation path planning should be constrained to the robot neighbourhood. The testing and analysis of the experimental results of trajectory tracking are reported. In this context, the different parameter weight performances of the cost function are studied. In this way, different kinds of trajectories are tested by using factor tuning.

Distributed model predictive control of irrigation canals

Irrigation canals are large-scale systems, consisting of many interacting components, and spanning vast geographical areas. For safe and efficient operation of these canals, maintaining the levels of the water flows close to pre-specified reference values is crucial, both under normal operating conditions as well as in extreme situations.    Irrigation canals are equipped with local controllers, to control the flow of water by adjusting the settings of control structures such as gates and pumps. Traditionally, the local controllers operate in a decentralized way in the sense that they use local information only, that they are not explicitly aware of the presence of other controllers or subsystems, and that no communication among them takes place. Hence, an obvious drawback of such a decentralized control scheme is that adequate performance at a system-wide level may be jeopardized, due to the unexpected and unanticipated interactions among the subsystems and the actions of the local controllers.    In this paper we survey the state-of-the-art literature on distributed control of water systems in general, and irrigation canals in particular. We focus on the model predictive control (MPC) strategy, which is a model-based control strategy in which prediction models are used in an optimization to determine optimal control inputs over a given horizon. We discuss how communication among local MPC controllers can be included to improve the performance of the overall system. We present a distributed control scheme in which each controller employs MPC to determine those actions that maintain water levels after disturbances close to pre-specified reference values. Using the presented scheme the local controllers cooperatively strive for obtaining the best system-wide performance. A simulation study on an irrigation canal with seven reaches illustrates the potential of the approach.

Smoothing Techniques for Distributed Model Predictive Control Algorithms in Networks*

Abstract In this paper we propose two dual decomposition methods based on smoothing techniques, called here the proximal center method and the interior-point Lagrangian method, to solve distributively separable convex problems. We show that some relevant centralized model predictive control (MPC) problems can be recast as a separable convex problem for which our dual methods can be applied. The new dual optimization methods are suitable for application to distributed MPC since they are highly parallelizable, each subsystem uses local information and the coordination between the local MPC controllers is performed via the Lagrange multipliers corresponding to the coupled dynamics or constraints.

Efficient Cooperative Distributed MPC using Partial Enumeration

We discuss in this paper a novel and efficient implementation of distributed Model Predictive Control (MPC) systems for large-scale systems. The method is based on Partial Enumeration (PE), an approach that allows to compute the (sub)optimal solution of the Quadratic Program associated to the MPC problem by using a solution table that stores only a few most recently optimal active sets. This method is applied to the each local MPC system with significant improvements in terms of computational efficiency, and the original PE algorithm is modified to guarantee robust stability of the overall closed-loop system. We also discuss how input constraints that involve different units, e.g. on the summation of common utility consumption, can be appropriately handled. We illustrate the benefits of proposed method by means a simulated example comprising three units.

Coordinating multiple optimization-based controllers: New opportunities and challenges

The status of using many, distributed optimization-based controllers for feedback control of large-scale, dynamic processes is presented and evaluated. We show that modeling the interactions between subsystems and exchanging trajectory information among subsystem model predictive controllers (MPCs) is insufficient to provide even closed-loop stability. The cause of this closed-loop instability is competition between the local agents. We next discuss the cooperative distributed MPC framework, in which the objective functions of the local MPCs are modified to achieve systemwide control objectives. This approach provides guaranteed nominal stability and performance properties, but at the cost of a high degree of communication between the local controllers. We next discuss the issue of taking advantage of the structure of the connections between the subsystems to reduce the required communication. The paper concludes by briefly presenting seven current and unsolved research challenges.

Coordinator MPC for maximizing plant throughput

Abstract In many cases, economic optimal operation is the same as maximum plant throughput, which is the same as maximum flow through the bottleneck(s). This insight may greatly simplify implementation. In this paper, we consider the case where the bottlenecks may move, with parallel flows that give rise to multiple bottlenecks and with crossover flows as extra degrees of freedom. With the assumption that the flow through the network is represented by a set of units with linear flow connections, the maximum throughput problem is then a linear programming (LP) problem. We propose to implement maximum throughput by using a coordinator model predictive controller (MPC). Use of MPC to solve the LP has the benefit of allowing for a coordinated dynamic implementation. The constraints for the coordinator MPC are the maximum flows through the individual units. These may change with time and a key idea is that they can be obtained with almost no extra effort using the models in the existing local MPCs. The coordinator MPC has been tested on a dynamic simulator for parts of the Karsto gas plant and performs well for the simulated challenges.

Coordinator MPC for maximizing plant throughput

In many cases, economic optimal operation is the same as maximum plant throughput, which is the same as maximum flow through the bottleneck(s). This insight may greatly simplify implementation. In this paper, we consider the case where the bottlenecks may move, with parallel flows that give rise to multiple bottlenecks and with crossover flows as extra degrees of freedom. With the assumption that the flow through the network is represented by a set of units with linear flow connections, the maximum throughput problem is then a linear programming (LP) problem. We propose to implement maximum throughput by using a coordinator model predictive controller (MPC). Use of MPC to solve the LP has the benefit of allowing for a coordinated dynamic implementation. The constraints for the coordinator MPC are the maximum flows through the individual units. These may change with time and a key idea is that they can be obtained with almost no extra effort using the models in the existing local MPCs. The coordinator MPC has been tested on a dynamic simulator for parts of the Kårstø gas plant and performs well for the simulated challenges.

COORDINATING MULTIPLE OPTIMIZATION-BASED CONTROLLERS: NEW OPPORTUNITIES AND CHALLENGES

The status of using many, distributed optimization-based controllers for feedback control of large-scale, dynamic processes is presented and evaluated. We show that modeling the interactions between subsystems and exchanging trajectory information among subsystem model predictive controllers (MPCs) is insufficient to provide even closed-loop stability. The cause of this closed-loop instability is competition between the local agents. We next discuss the cooperative distributed MPC framework, in which the objective functions of the local MPCs are modified to achieve systemwide control objectives. This approach provides guaranteed nominal stability and performance properties, but at the cost of a high degree of communication between the local controllers. We next discuss the issue of taking advantage of the structure of the connections between the subsystems to reduce the required communication. The paper concludes by briefly presenting seven fcurrent and unsolved research challenges.

Predictive Function Control Based on Multi-model for pH Plant

A multi-model structure for predictive function control has been presented for a pH neutralisation process. The non linear pH process is partitioned into several local operating region. Within each region, a local predictive function controller is constructed based on the corresponding local linear model, their outputs are combined to form a global control action by using their membership functions. i.e. the global predictive function controller (GC-PFC) consists of a set of local model predictive function controller (LM-PFC) by weighting. The GC-PFC has been applied to quality control of a pH non linear process. Realistic simulation shows that GC-PFC is an attractive option for non linear MPC.