Data-based Predictive Control Research Articles

Data-based distributed model predictive control for large-scale systems

Data-based distributed model predictive control for large-scale systems

Data-driven adaptive predictive frequency control for power systems with unknown and time-varying inertia

This paper proposes a novel data-based adaptive predictive frequency control method for multi-area power systems with unknown and time-varying inertia. Firstly, a data-based representation is built and updated at each instant based on behavioural system theory by using historical input–output data, where a moving horizon estimation method is used to deal with the unknown time-varying inertia issue adaptively. Then, the optimal frequency control signal is computed by solving an optimization problem under the framework of data-based predictive control. Simulation results on a power system with three control areas demonstrate the effectiveness of the proposed method.

A Direct Optimal Input Determination Data-Based Predictive Current Control for PMSM Drives Without System Identification

A Direct Optimal Input Determination Data-Based Predictive Current Control for PMSM Drives Without System Identification

Data-based adaptive predictive bipartite consensus for nonlinear multiagent systems against DoS attacks

This article investigates a Denial-of-Service (DoS) attack problem for nonlinear unknown discrete-time multiagent systems (MASs) to implement bipartite consensus tracking tasks with fixed and switching topologies. Firstly, an equivalent linearization data model of each agent is constructed using a pseudo partial derivative approach, where only one parameter needs to be estimated using input/output data of the controlled MASs. Meanwhile, the DoS attack behavior is described by a Bernoulli distribution process, and both cooperative and competitive relationships among agents are investigated. Moreover, an increment prediction compensator is designed to reduce the effect of DoS attacks. A data-based adaptive predictive bipartite consensus control algorithm is formulated. The corresponding theoretical analysis indicates that tracking errors of MASs with fixed and switching topologies converge to a small range around zero. Finally, several simulations and hardware tests further verify the proposed scheme’s effectiveness.

Development of data-based model predictive control for continuous damping and air spring suspension system

This paper presents the development of a data-based model predictive control method for a semi-active suspension system with air springs and continuous damping. A continuous damping controller (CDC) has been devised for the system to alter its damping coefficient in real-time, capable of reducing the impact of external road disturbances. In this research, the damping force has been split up into a nominal force and a controllable additional force, allowing the system to be modelled as a linear time-invariant system, despite the inherent nonlinearity of air spring suspension systems. In consideration of such constraints given by the damper, a model predictive controller (MPC) has been devised with the goal of improving ride comfort. Additionally, Gaussian process regression (GPR) has been used to compensate for output estimation errors arising from model parameter uncertainties. The semi-active suspension system also featured a multichamber air spring with three available modes of stiffness. Hence, a stiffness controller has been designed to select an appropriate mode based on the predicted vehicle states given by the MPC using road preview information and a reduced full-car model. The proposed algorithm has been verified using computer simulations. The results showed that compensations for model errors made with GPR significantly improves ride comfort even in the presence of parameter uncertainties. Additionally, both the damping controller and stiffness mode selector were successfully implemented into an actual test vehicle. Vehicle test results showed the proposed algorithm to be robust and effective in enhancing ride comfort and reducing vehicle pitch motion.

Robust data-based predictive control of systems with parametric uncertainties: Paving the way for cooperative learning

This article combines data and tube-based predictive control to deal with systems with bounded parametric uncertainty. This integration generates robustly feasible control sequences that can also be exploited in cooperative scenarios where controllers learn from each other’s data. In particular, the approach is based on a database that contains information from previous executions of the same and other controllers handling similar systems. By the combination of feasible histories plus an auxiliary control law that deals with bounded uncertainties, which only needs to be stabilizing for at least one of the system realizations within the uncertainty set, this scheme provides a finite-horizon predictive controller that guarantees exponential stability and robust constraint satisfaction. The validity and benefits of the proposed scheme are shown in case studies with linear and non-linear dynamics.

Data-Based Predictive Control for Power Congestion Management in Subtransmission Grids Under Uncertainty

Data-Based Predictive Control for Power Congestion Management in Subtransmission Grids Under Uncertainty

Optimal management and data-based predictive control of district heating systems: The Novate Milanese experimental case-study

This paper addresses the modelling and optimal control of district heating systems connected to the electrical grid, with the goal of maximizing their operational efficiency and enabling the participation to electricity markets. Being these systems governed by nonlinear large-scale dynamical models, a novel procedure is proposed, which enables to obtain suitable models for optimization, and consisting in a combination of physical and identified piece-wise linear models. A two-phase optimization and control scheme is then designed, including an offline scheduling problem for participating to the day-ahead energy market, and an online Model Predictive Control system, minimizing the energy consumption of thermal generators while properly satisfying the users thermal demand. The proposed methodology is developed considering a real district heating plant, owned by the energy company A2A S.p.A. and supplying the city of Novate Milanese (Italy), and different experiments on the plant have been carried out. The experimental results and achieved performances are promising, showing a significant reduction of the operational costs and overall gas consumption.

Towards reliable data-based optimal and predictive control using extended DMD

While Koopman-based techniques like extended Dynamic Mode Decomposition are nowadays ubiquitous in the data-driven approximation of dynamical systems, quantitative error estimates were only recently established. To this end, both sources of error resulting from a finite dictionary and only finitely-many data points in the generation of the surrogate model have to be taken into account. We generalize the rigorous analysis of the approximation error to the control setting while simultaneously reducing the impact of the curse of dimensionality by using a recently proposed bilinear approach. In particular, we establish uniform bounds on the approximation error of state-dependent quantities like constraints or a performance index enabling data-based optimal and predictive control with guarantees.

Scenario Model Predictive Control for Data-Based Energy Management in Plug-In Hybrid Electric Vehicles

One of the major limitations of optimization-based strategies for allocating the power flow in hybrid powertrains is that they rely on predictions of future power demand. These predictions are inherently uncertain as they are dependent on complex human behaviours that are challenging to model accurately. This paper proposes a data-based scenario model predictive control framework, where the inputs are determined at each control update by optimizing the power allocation over multiple previous examples of a route being driven. The proposed energy management optimization is convex, and results from scenario optimization are used to bound the confidence that the one-step-ahead optimization will be feasible with given probability. It is shown through numerical simulation that scenario model predictive control (MPC) obtains the same reduction in fuel consumption as nominal MPC with full preview of future driver behaviour, and that the scenario MPC optimization can be solved efficiently using a tailored optimization algorithm.

Data-based robust model predictive control for wastewater treatment process

Data-driven methods are widely used in wastewater treatment processes (WWTPs). However, disturbances are not fully considered during the control process, which may affect the stable operation of WWTPs. To address this issue, a data-driven robust model predictive control (DRMPC) method is proposed for WWTPs with disturbances. First, an interval type-2 fuzzy neural network (IT2FNN), which has high robust performance to disturbances, is used to identify dynamic behavior of WWTPs. Second, based on IT2FNN, a robust stabilizing controller is developed to weaken the influence of disturbances on control performance and stability. Third, the proposed controller ensures constraint satisfaction and input-to-state stability (ISS). Finally, the feasibility of DRMPC strategy is verified on the benchmark simulation model No. 1 (BSM1). The experimental studies indicate that the proposed DRMPC is capable of running stably and tracking accurately.

Data-Based Receding Horizon Control of Linear Network Systems

We propose a distributed data-based predictive control scheme to stabilize a network system described by linear dynamics. Agents cooperate to predict the future system evolution without knowledge of the dynamics, relying instead on learning a data-based representation from a single sample trajectory. We employ this representation to reformulate the finite-horizon Linear Quadratic Regulator problem as a network optimization with separable objective functions and locally expressible constraints. We show that the controller resulting from approximately solving this problem using a distributed optimization algorithm in a receding horizon manner is stabilizing. We validate our results through numerical simulations.

Model predictive air path control for a two-stage turbocharged spark-ignition engine with low pressure exhaust gas recirculation

Innovative air path concepts for turbocharged spark-ignition engines with exhaust gas recirculation impose high demands on the control due to nonlinearities and cross-couplings. This contribution investigates the control of the air and exhaust gas recirculation paths of a two-stage turbocharged spark-ignition engine with low pressure exhaust gas recirculation. Using exhaust gas recirculation at high loads, the in-cylinder temperature can be lowered, reducing the knock tendency, while at the same time preventing the need for the enrichment of the air/fuel ratio. Air and exhaust gas recirculation paths are cross-coupled and show different delay times. To tackle these challenges, a data-based two-stage model predictive controller is proposed: The target selector accounts for the overactuated system structure, while the dynamic controller adjusts the charging pressure and exhaust gas recirculation rate. The prediction model setup is based on a small amount of dyno-run measurement data. To ensure real-time capability, the model is kept as simple as possible. This allows for fast turnaround times of the algorithm, while maintaining the necessary accuracy in steady-state and transient operation. This study focuses on a two-stage control concept based on a target selector for optimal stationary control inputs and the dynamic controller considering the dynamic behavior of the air and exhaust gas recirculation paths. Subsequently, the control concept for the two-stage turbocharged spark-ignition engine with low pressure exhaust gas recirculation is validated via experimental tests under real-driving conditions on an automotive test track, using a prototype test vehicle. Results show that boost pressure as well as exhaust gas recirculation rate setpoints are met without overshoot and control deviation with settling times being close to the boundaries set by the hardware.

Output feedback MPC based on smoothed projected kinky inference

In this study, the authors propose a stabilising data-based model predictive controller for systems subject to constraints in which the prediction model is inferred from experimental data of the plant using a machine learning technique. The inference method is a modification of the kinky inference tailored for model predictive control. In particular, the modified method has a lower computational effort and provides smoother predictions than the original method. The controller formulation considers soft constraints in the outputs, hard constraints in the inputs and guarantees closed-loop robust stability as well as performance by means of the use of different control and prediction horizons and a weighted terminal cost. Under the assumption that the model of the system is Hölder continuous, they prove that the closed-loop system is input-to-state stable with respect to the estimation errors. The results are demonstrated in a case study of a continuously stirred-tank reactor.

Data‐based predictive control for networked non‐linear multi‐agent systems consensus tracking via cloud computing

This study investigates the consensus tracking problem for a class of networked non-linear multi-agent systems (NNMASs) using cloud computing. To achieve the stability and output consensus of the NNMAS and to actively compensate for two-channel network delays, a data-based cloud predictive control scheme is proposed. The design of the proposed novel control scheme, which only depends on the historical input and output data of the agents without using the explicit or implicit information of its structure, is detailed. Sufficient conditions are derived to guarantee the stability and output consensus of the closed-loop NNMAS. Both numerical simulations and cloud-based practical experiments are conducted to demonstrate the effectiveness of the proposed scheme. The outcome promotes the engineering application of cloud computing in multi-agent systems.

Data-Based Predictive Control for Wastewater Treatment Process

Wastewater treatment process (WWTP) has long been a challenging industrial issue due to its built-in uncertainties and discontinuous measurement of system states. To solve this problem, in this paper, a data-based predictive control (DPC) strategy, based on the available sensing measurements, is proposed to control the dissolved oxygen (DO) concentration in WWTP. First, a self-organizing fuzzy neural network, which can adjust both the structure and parameters simultaneously, is developed to identify the real-time states of WWTP. Second, an improved nonlinear predictive control method is designed to reduce the online computation complexity by transforming the constrained conditions into an unconstrained nonlinear programming problem. Then, an adaptive second-order Levenberg-Marquardt algorithm is developed to derive the control law of DPC. Third, the theoretical analysis on the stability is also given to confirm the prerequisite of any successful application of DPC. Finally, the proposed DPC strategy is applied to the Benchmark Simulation Model No. 1. Experimental results demonstrate that the control performance of the proposed DPC is better than some existing methods.

Data-based predictive control via direct weight optimization

Abstract In this paper we propose a novel data-based predictive control scheme in which the prediction model is obtained from a linear combination of past system trajectories. The proposed controller optimizes the weights of this linear combination taking into account simultaneously performance and the variance of the estimation error. For unconstrained systems, dynamic programming is used to obtain an explicit linear solution of a finite or infinite horizon optimal control problem. When constraints are taken into account, the controller needs to solve online a quadratic optimization problem to obtain the optimal weights, possibly considering also local information to improve the performance and estimation. A simulation example of the application of the proposed controller to a quadruple-tank system is provided.

Data-based predictive control for networked nonlinear systems with packet dropout and measurement noise

In this paper, the data-based control problem is investigated for a class of networked nonlinear systems with measurement noise as well as packet dropouts in the feedback and forward channels. The measurement noise and the number of consecutive packet dropouts in both channels are assumed to be random but bounded. A data-based networked predictive control method is proposed, in which a sequence of control increment predictions are calculated in the controller based on the measured output error, and based on the control increment predictions received by the actuator, a proper control action is obtained and applied to the plant according to the real-time number of consecutive packet dropouts at each sampling instant. Then the stability analysis is performed for the networked closed-loop system. Finally, the effectiveness of the proposed method is illustrated by a numerical example.

Loss-of-Control Mitigation via Predictive Cuing

Flying near the edge of the safe operating envelope is an inherently unsafe proposition. The edge of the envelope here implies that small changes or disturbances in the system state or system dynamics can take the system out of the safe envelope in a short time and could result in loss-of-control events. This study evaluates approaches to pilot cueing based on predicting loss-of-control safety margins as the aircraft gets closer to the edge of the safe operating envelope. The goal is to provide the pilot aural, visual, and tactile cues focused on maintaining the pilot’s control action within predicted loss-of-control boundaries. The predictive architecture presented in this paper combines quantitative loss-of-control boundaries, an adaptive prediction, and data-based predictive control algorithms. The combined architecture is applied to a nonlinear transport-class aircraft. Evaluations of various loss-of-control cues using both test and commercial pilots in the NASA Ames Research Center’s vertical motion-base simulator were conducted in the summer of 2013. The paper presents details of the cueing algorithms and the results of the evaluation focused on effectiveness of these cues in preventing the pilots from entering a loss-of-control event.

Data-Based Predictive Control for Networked Nonlinear Systems With Network-Induced Delay and Packet Dropout

This paper addresses the data-based networked control problem for a class of nonlinear systems. Network communication constraints, such as network-induced delay, packet disorder, and packet dropout in both the feedback and forward channels, are considered and further treated as the round-trip time (RTT) delay that is redefined. By using the packet-based transmission mechanism and the model-free adaptive control algorithm, a data-based networked predictive control method is proposed to actively compensate for the random RTT delay. The proposed method does not require any information on the plant model and depends only on the input and output data of the plant. A simple and explicit sufficient condition, which is related to the upper bound of the RTT delays, is derived for the stability of the closed-loop system. Additionally, a zero steady-state output tracking error can be achieved for a step reference input. The effectiveness of the proposed method is demonstrated via simulation and experimental results.