Robust model predictive control for continuous nonlinear systems with the quasi-infinite adaptive horizon algorithm

This work introduces an adaptive model predictive control (MPC) scheme for nonlinear processes. It is well-known that the conventional MPC uses linear model to represent local dynamics of a nonlinear process and resulting optimization problem is convex. However, the model mismatch between linear model and nonlinear process may significantly degrade the performance of MPC and thereby the closed-loop stability may not be guaranteed. To address this issue, the model used in MPC should be accurate enough for different operating conditions. The contribution of this paper is the development of a steady state constrained autoregressive with exogenous terms model (SSARX) that is still linear and adaptive to the variation of operating conditions. This new model then can be integrated with a linear MPC to achieve a trade-off between computational complexity and model accuracy. The proposed controller is applied in a continuous fermentation process and compared to an MPC with unconstrained ARX model (non-adaptive). The simulation shows that the MPC with ARX model leads to an offset during setpoint tracking while the MPC with SSARX model is offset free.

Pitch control is related to the efficient energy conversion and safe operation of wind turbine, which is a challenging control issue because of the highly nonlinear dynamics of wind turbine, system constraints and stochastic environmental disturbances, etc. In this article, we propose the adaptive model predictive pitch control strategy based on linear parameter varying (LPV) model to solve the pitch control issue of large-scale wind turbine. Firstly, a high-fidelity LPV model is constructed to approximate the high-nonlinearity dynamics of wind turbine. The gap metric theory is introduced to optimize the selection of local models. Based on the proposed LPV model, an adaptive model predictive controller with linear time-varying Kalman filter is designed for large-scale wind turbine. The benchmark 5 megawatt (MW) wind turbine model provided by FAST is employed to evaluate the controller performance. The results show that, compared with the traditional gain scheduling proportional–integral control strategy, the proposed control strategy can achieve better power regulation and load reduction performance under gust, step and turbulent wind.

Adaptive MPC for Uncertain Discrete-Time LTI MIMO Systems with Incremental Input Constraints

In this article, we present a quasi-infinite horizon nonlinear model predictive control (MPC) scheme for tracking of generic reference trajectories. This scheme is applicable to nonlinear systems, which are locally incrementally stabilizable. For such systems, we provide a reference generic offline procedure to compute an incrementally stabilizing feedback with a continuously parameterized quadratic quasi-infinite horizon terminal cost. As a result, we get a nonlinear reference tracking MPC scheme with a valid terminal cost for general reachable reference trajectories without increasing the online computational complexity. The practicality of this approach is demonstrated with a benchmark example.

This paper aims to design a stabilized Adaptive Model Predictive Control (MPC) for nonlinear continuous systems with unknown dynamics at the start time. The proposed method is based on the Lyapunov MPC (LMPC) where the derivative of the Lyapunov function of the closed-loop system is limited to be less than that under an auxiliary Lyapunov controller. Different from the existing LMPC methods, this paper provides an online constraint updating strategy, which gradually reduces the conservation of the constraint. Specifically, the piece-wise linear (PWL) model updated by the Set-Membership (SM) identification method is adopted in the constraint and the auxiliary controller design, which can result in a non-increasing boundary of the model error. Then, a sufficient condition that guarantees the stabilization of the closed-loop system is deduced, which takes the effect of sample-and-hold implementation and the error boundary of the PWL model into account. By this condition, an optimization problem to update the Lyapunov controller for relaxing the constraints of MPC is proposed. We prove that, by the proposed method, the states will eventually converge to a small region around the equilibrium, both this small region and the conservatism of MPC decrease with the increasing of the accuracy of the PWL model. The application of the proposed method to a chemical process demonstrates the effectiveness of the proposed method. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —The control problem of nonlinear continuous systems with unknown dynamics has attracted the attention of researchers. This article develops an Adaptive MPC for this kind of system, which updates the boundary of uncertainties according to the measurement, and updates the constraints of MPC according to the updated uncertainty boundary. It provides a guaranteed stabilization and continuously relaxed constraint design for MPC where machine learning-based models are employed to predict the future trajectory. In the process of implementation, the method solves two optimization problems: one is to get the parameters of constraints which is based on the PWL model, and the other is used to optimize the future trajectory with the updated constraints. To proceed, the initial boundary of the uncertainty should be given at the initial instant, the initialization of PWL models and constraints can be configured based on it. The proposed method can be used for tracking control of chemical processes, power systems, robots, etc., where the accurate dynamics are not provided in advance and an improved performance is required.

In this article, a nonlinear disturbance observer‐based multiple‐model adaptive explicit model predictive control (NDO‐MMAEMPC) scheme is developed for a nonlinear MIMO system with parametric uncertainty as well as an external disturbance. The proposed method manages external disturbances without affecting the degree of smoothness of the control signals. Further, to cope with the unknown system uncertainty, blending‐based adaptive identification schemes are used for the same class of systems. For each identification model, an explicit nonlinear model predictive controller is computed off‐line for the corresponding model in advance, which saves computation power during operation. The generated control inputs from the set of explicit controllers are being blended online using adaptive weight. An extended Kalman filter algorithm is employed for the estimation of the unavailable states of the nonlinear system. Using an aerodynamic laboratory setup, the effectiveness of the proposed control algorithm is verified.

Robustly stable adaptive horizon nonlinear model predictive control

In the navigation of underwater robots, large ocean current disturbance often causes significant tracking errors. To better resist ocean current disturbance, the hydrodynamic characteristics of the spherical underwater robot are studied, and a model predictive control strategy based on adaptive model parameters is proposed, according to these characteristics. Firstly, the hydrodynamic characteristics of the robot under static water and constant flow disturbance were obtained and analyzed by the computational fluid dynamics method. Then, the dynamic models of the robot under different disturbances could be calculated from the data obtained, based on the least square method. Finally, an adaptive model predictive control (AMPC) strategy, with an ocean current observer, was designed, based on the dynamic models. When the current disturbance velocity was twice the robot velocity, the proposed strategy reduced the tracking error by 39% and 42% in X and Y directions, respectively. In addition, the hydrodynamic characteristics were verified by experiments.

This article presents an optimization-based control framework for the autonomous forced landing of a fixed-wing unmanned aircraft (UA). A two-level model-predictive control (MPC) scheme is proposed to realize this framework, where an economic model-predictive control (EMPC) in a long piecewise constant fashion is proposed at the high level, while a short fixed-horizon linear time-varying MPC at the low level responds to fast dynamics of UA and tracks the optimal path provided by the high-level controller, alleviating computational burden compared to the high-frequency single-layer MPC scheme. Compared with a single EMPC setup with high sampling frequency, this hierarchical EMPC controller can significantly reduce the computational complexity and make it feasible to be implemented in real time. In addition, it also responds to disturbances (e.g., wind) and aircraft fast dynamics in a timely manner. The recursive feasibility and stability of the high- and low-level MPC schemes are established. The performance of the proposed EMPC forced landing function is illustrated by simulation case studies on both Aerosonde and Skywalker X8, compared favorably with competing schemes.

Discussion on: “Improved MPC Design based on Saturating Control Laws”

In this paper, we present a data‐driven model predictive control (MPC) scheme that is capable of stabilizing unknown linear time‐invariant systems under the influence of process disturbances. To this end, Willems' lemma is used to predict the future behavior of the system. This allows the entire scheme to be set up using only a priori measured data and knowledge of an upper bound on the system order. First, we develop a state‐feedback MPC scheme, based on input‐state data, which guarantees closed‐loop practical exponential stability and recursive feasibility as well as closed‐loop constraint satisfaction. The scheme is extended by a suitable constraint tightening, which can also be constructed using only data. In order to control a priori unstable systems, the presented scheme contains a prestabilizing controller and an associated input constraint tightening. We first present the proposed data‐driven MPC scheme for the case of full state measurements, and also provide extensions for obtaining similar closed‐loop guarantees in case of output feedback. The presented scheme is applied to a numerical example.

The efficiency of a SAGD operation depends on developing a uniform steam chamber and maintaining an optimal subcool along the length of the well pair. Heterogeneity in reservoir properties may lead to suboptimal subcool levels. Recently, Model Predictive Control (MPC) based on the real-time production, temperature, and pressure data along with other well and surface constraint information has been proposed for a real-time feedback control of SAGD well pairs. Reservoir dynamics in MPC is represented using either linear step response model or one-dimensional ordinary differential equation. However, such simplified models are insufficient in MPC since SAGD is more complex and highly nonlinear process. Therefore, MPC framework that represents nonlinear behaviour of SAGD over an extended control period is required to achieve optimized subcool and steam conformance. In this research, two novel workflows are proposed to handle nonlinear reservoir dynamics in MPC. First approach known as Adaptive MPC includes recursive estimations at each control interval based on system identification theory. This allows evolution of the coefficients of a fixed model structure such that the updated system identification model in MPC controller reflects current reservoir dynamics adequately. Another approach, Gain-Scheduled MPC, decomposes the subcool control problem in a parallel manner and uses a bank of multiple controllers rather than only one controller. This ensures effective control of the nonlinear reservoir system even in adverse control situations by employing aggressive variations in input parameters. Suggested workflows are implemented using history-matched numerical model of a reservoir located in northern Alberta. Steam injection rates and liquid production rate are considered as input variables in MPC, constrained to available surface facilities. Well-pair is divided into multiple sections and subcool of each section is considered as an output variable. Optimum set-point for subcool is considered as 20°C. Results are compared with actual field data (in which no control algorithm is used) and analyzed based on two criteria: 1) Do all subcools track optimum set-point while maintaining stability in input variables and 2) Does net present value (NPV) of oil improve in case of Adaptive and Gain-Scheduled MPC? In general, we conclude that both Adaptive and Gain-Scheduled MPC provide superior tracking of subcool set-point and hence better steam conformance due to adequate representation of reservoir dynamics by recursive estimation of coefficients and multiple controllers. In addition, results indicate stability in input parameters and improvement in economic performance. NPV is improved by 23.69% and 10.36% in case of Adaptive and Gain-Scheduled MPC, respectively. Under current economic scenario, proposed workflows can improve the NPV of a SAGD reservoir by optimizing the well operational parameters while considering constraints of surface facilities and minimizing environmental footprints.

In this paper, a “third generation” benchmark problem that focuses on the control of wind excited response of a tall building, using the Model Predictive Control (MPC) scheme, is presented. A 76 story, 306 m tall concrete office tower proposed for the city of Melbourne, Australia, is being used to demonstrate the effectiveness of MPC. The MPC scheme is based on an explicit use of a prediction model of the system response to obtain the control actions by minimizing an objective function. Optimization objectives in MPC include minimization of the difference between the predicted and desired response trajectories, and the control effort subjected to prescribed constraints. By incorporating input/output hard constraints, the MPC scheme provides an optimal control force that satisfies the prescribed constraints.

A model predictive control for a multi-chiller system in data center considering whole system energy conservation

Modeling and model predictive control of the BioPower combined heat and power (CHP) plant