Data-driven Predictive Control Research Articles

Data-driven predictive control of Hammerstein–Wiener systems based on subspace identification

It poses significant challenge to control Hammerstein–Wiener systems involving modeling nonlinearities. In this paper, a novel data-driven predictive control method based on the subspace identification of Hammerstein–Wiener systems is presented. By reformulating the open- and closed-loop Hammerstein–Wiener model, subspace predictions of the outputs are derived using recursive substitution of the Hankel matrices. The output nonlinearity is presented by polynomial representation and the subspace predictors are obtained using the QR decomposition, together with additional algebra manipulations, where Q is an orthogonal matrix and R is an upper triangular matrix. The predictors are applied to the model predictive controller, wherein the integrated action is successfully incorporated. The effectiveness and feasibility of the proposed controller is also verified by numerical simulation on a fermentation bioreactor system.

Data-Driven Predictive Control with Nonlinear Compensation for Performance Management in Virtualized Software System

This paper deals with data-driven predictive control for relative performance management in virtualized software system. The system dynamics are characterized in Hammerstein-Wiener structure to capture nonlinear and linear characteristics. The proposed control approach is the implementation of Subspace-based Predictive Control with the integration of nonlinear compensation. The compensator functions are inverse static input and output nonlinearity models from the Hammerstein-Wiener system identification. The subspace predictors are formulated from the linear model input and output of Wiener block. The experimental results from three scenarios of performance objectives show the reliability of Subspace-based Predictive Control to manage the virtualized software system.

Data-Driven Predictive Torque Coordination Control during Mode Transition Process of Hybrid Electric Vehicles

Torque coordination control significantly affects the mode transition quality during the mode transition dynamic process of hybrid electric vehicles (HEV). Most of the existing torque coordination control methods are based on the mechanism model, whose control effect heavily depends on the modeling accuracy of the HEV powertrain. However, the powertrain structure is so complex, that it is difficult to establish its precise mechanism model. In this paper, a torque coordination control strategy using the data-driven predictive control (DDPC) technique is proposed to overcome the shortcomings of mechanism model-based control methods for a clutch-enabled HEV. The proposed control strategy is only based on the measured input-output data in the HEV powertrain, and no mechanism model is needed. The conflicting control requirements of comfortability and economy are included in the cost function. The actual physical constraints of actuators are also explicitly taken into account in the solving process of the data-driven predictive controller. The co-simulation results in Cruise and Simulink validate the effectiveness of the proposed control strategy and demonstrate that the DDPC method can achieve less vehicle jerk, faster mode transition and smaller clutch frictional losses compared with the traditional model predictive control (MPC) method.

Data-driven predictive control for continuous-time linear parameter varying systems with application to wind turbine

A new data-driven predictive control method based on subspace identification for continuous-time linear parameter varying (LPV) systems is presented in this paper. It is developed by reformulating the continuous-time LPV system which utilizes Laguerre filters to obtain the subspace prediction of output. The subspace predictors are derived by QR decomposition of input-output and Laguerre matrices obtained by input-output data. The predictors are then applied to design the model predictive controller. It is shown that the integrated action is incorporated in the control effect to eliminate the steady-state offset. We control the continuous-time LPV systems to obtain the attractive performance with the proposed data-driven predictive control method. The proposed controller is applied to a wind turbine to verify its effectiveness and feasibility.

Active queue management algorithm based on data-driven predictive control

Model predictive control (MPC) is a popular strategy for active queue management (AQM) that is able to incorporate physical and user defined constraints. However, the current MPC methods rely on explicit fluid model of TCP behavior with input time delay. In this paper, we propose a novel AQM algorithm based on data-driven predictive control, called Data-AQM. For Internet system with large delay, complex change and bad disturbance, data-driven predictive controller can be obtained directly based on the input---output data alone and does not require any explicit model of the system. According to the input---output data, the future queue length in data buffer, which is the basis of optimizing drop probability, is predicted. Furthermore, considering system constraints, the control requirement is converted to the optimal control objective, then the drop probability is obtained by solving the optimal problem online. Finally, the performances of Data-AQM are evaluated through a series of simulations.

Data-driven Predictive Control of Micro Gas Turbine Combined Cooling Heating and Power system

Abstract: Micro gas turbine-based combined cooling, heating and power (MGT-CCHP) system is in an important direction toward the development of smart buildings and district energy systems, which provides a clean, highly efficient and reliable means of producing energy for multiple use. However, the control of MGT-CCHP system is a challenge, due to its behavior such as large thermal inertia, and strong coupling among multi-variables. For this reasons, this paper develops a data-driven predictive controller for the MGT-CCHP system to improve its operating performance. The technique of subspace identification is utilized to construct the predictor directly from the input-output data, which can be used to estimate the future behavior of the system. The predictive controller is then designed to regulate the multi-variable MGT-CCHP system under the input-constraints. The effectiveness of the proposed control approach is demonstrated through simulation results on an 80kw MGT-CCHP simulator.

Data-driven predictive iterative learning control for a class of multiple-input and multiple-output nonlinear systems

This paper presents a novel data-driven predictive iterative learning control (DDPILC) scheme based on a new dynamic linearization technique along the iteration axis for a class of repeatable multiple-input and multiple-output nonlinear discrete-time systems. The proposed DDPILC scheme combines iterative learning control with predictive control and the distinct feature of the scheme is that the controller design depends only on the measured input/output data without using any model information of the controlled plant. In addition, if the control system is subjected to input and output constraints, the constrained DDPILC scheme is also proposed. The theoretical analysis shows that with random initial operation conditions, the proposed unconstrained DDPILC scheme guarantees monotonic and pointwise convergence while the constrained DDPILC scheme guarantees the asymptotic and pointwise convergence. The applicability and effectiveness of the proposed DDPILC schemes are further verified through simulations.

Data-Driven Predictive Gearshift Control for Dual-Clutch Transmissions and FPGA Implementation

During the gearshift management process, the shift time and the shift shock affect the shift quality (smoothness and efficiency) greatly. In this paper, a gearshift controller is designed using a data-driven predictive control technique to improve the shift quality of vehicles with a dual-clutch transmission (DCT). It is directly obtained from the input-output data of a DCT model constructed by the commercial software AMESim. In order to obtain an offset-free control for the reference input, a predictor equation is gained with the incremental input and outputs. The conflicting control requirements of a short shift time and a small shift shock are taken into account in the optimal objective function by smoothly tracking a selected proper reference trajectory and limiting the change rate of the actuator. Moreover, due to physical characteristics, the constraints of the actuator are explicitly considered in the problem formulation. Finally, the effectiveness of the proposed data-driven predictive controller is tested in the AMESim simulation model of a DCT vehicle. Aiming to pursue a real-time improvement of the hardware computing speed for the data-driven predictive controller, a novel scheme for implementing the designed controller based on a field-programmable gate array (FPGA) is proposed. Moreover, a rapid prototyping platform based on an FPGA and dSPACE is introduced for testing and verifying the computational performance of the proposed controller. The experimental results show that the designed gearshift control algorithm for gearshift has better performance in real time, and the control performance can be guaranteed in a real-time environment.

Adaptive Predictive Control: A Data-Driven Closed-Loop Subspace Identification Approach

This paper presents a data-driven adaptive predictive control method using closed-loop subspace identification. As the predictor is the key element of the predictive controller, we propose to derive such predictor based on the subspace matrices which are obtained through the closed-loop subspace identification algorithm driven by input-output data. Taking advantage of transformational system model, the closed-loop data is effectively processed in this subspace algorithm. By combining the merits of receding window and recursive identification methods, an adaptive mechanism for online updating subspace matrices is given. Further, the data inspection strategy is introduced to eliminate the negative impact of the harmful (or useless) data on the system performance. The problems of online excitation data inaccuracy and closed-loop identification in adaptive control are well solved in the proposed method. Simulation results show the efficiency of this method.

A Novel Data-Driven Terminal Iterative Learning Control with Iteration Prediction Algorithm for a Class of Discrete-Time Nonlinear Systems

A data-driven predictive terminal iterative learning control (DDPTILC) approach is proposed for discrete-time nonlinear systems with terminal tracking tasks, where only the terminal output tracking error instead of entire output trajectory tracking error is available. The proposed DDPTILC scheme consists of an iterative learning control law, an iterative parameter estimation law, and an iterative parameter prediction law. If the partial derivative of the controlled system with respect to control input is bounded, then the proposed control approach guarantees the terminal tracking error convergence. Furthermore, the control performance is improved by using more information of predictive terminal outputs, which are predicted along the iteration axis and used to update the control law and estimation law. Rigorous analysis shows the monotonic convergence and bounded input and bounded output (BIBO) stability of the DDPTILC. In addition, extensive simulations are provided to show the applicability and effectiveness of the proposed approach.

Predictive pre-cooling of thermo-active building systems with low-lift chillers

This article describes the development and experimental validation of a data-driven model predictive control algorithm that optimizes the operation of a low-lift chiller, a variable-capacity chiller run at low pressure ratios, serving a single zone with a thermo-active building system. The predictive control algorithm incorporates new elements lacking in previous chiller pre-cooling control optimization methods, including a model of temperature and load-dependent chiller performance extending to low-pressure and part-load ratios and a data-driven zone temperature response model that accounts for the transient thermal response of a concrete-core radiant floor thermo-active building system. Data-driven models of zone and concrete-core thermal response are identified from monitored zone temperature and thermal load data and combined with an empirical model of a low-lift chiller to implement model predictive control. The energy consumption of the cooling system, including the chiller compressor, condenser fan, and chilled-water pump energy, is minimized over a 24-h look-ahead moving horizon using the thermo-active building system for thermal storage and radiant distribution. A generalized pattern-search optimization over compressor speed is performed to identify optimal chiller control schedules at every hour, thereby accomplishing load shifting, efficient part-load operation, and cooling energy savings. Results from testing the system's sensible cooling efficiency in an experimental test chamber subject to the typical summer week of two climates, Atlanta, GA, and Phoenix, AZ, show sensible cooling energy savings of 25% and 19%, respectively, relative to a high efficiency, variable-speed split-system air conditioner.

Data-driven predictive control for networked control systems

This paper is concerned with the problem of data-driven predictive control for networked control systems (NCSs), which is designed by applying the subspace matrices technique, obtained directly from the input/output data transferred from networks. The networked predictive control consists of the control prediction generator and network delay compensator. The control prediction generator provides a set of future control predictions to make the closed-loop system achieve the desired control performance and the network delay compensator eliminates the effects of the network transmission delay. The effectiveness and superiority of the proposed method is demonstrated in simulation as well as experiment study.

Design of a Data-Driven Predictive Controller for Start-up Process of AMT Vehicles

In this paper, a data-driven predictive controller is designed for the start-up process of vehicles with automated manual transmissions (AMTs). It is obtained directly from the input-output data of a driveline simulation model constructed by the commercial software AMESim. In order to obtain offset-free control for the reference input, the predictor equation is gained with incremental inputs and outputs. Because of the physical characteristics, the input and output constraints are considered explicitly in the problem formulation. The contradictory requirements of less friction losses and less driveline shock are included in the objective function. The designed controller is tested under nominal conditions and changed conditions. The simulation results show that, during the start-up process, the AMT clutch with the proposed controller works very well, and the process meets the control objectives: fast clutch lockup time, small friction losses, and the preservation of driver comfort, i.e., smooth acceleration of the vehicle. At the same time, the closed-loop system has the ability to reject uncertainties, such as the vehicle mass and road grade.

A data-driven bilinear predictive controller design based on subspace method

In this paper, a new data-driven model predictive control (MPC), based on bilinear subspace identification, is considered. The system's nonlinear behavior is described with a bilinear subspace predictor structure in an MPC framework. Thus, the MPC formulation results in a fixed structure objective function with constraints regardless of the underlying nonlinearity. For unconstrained systems, the identified subspace predictor matrices can be directly used as controller parameters. Therefore, we design optimization algorithms that exploit this feature. The open-loop optimization problem of MPC that is nonlinear in nature is solved with series quadratic programming (SQP) without any approximations. The computational efficiency already demonstrated with the current formulation presents further opportunities to enable online control of nonlinear systems. These improvements and close integration of modeling and control also eliminate the intermediate design step, which provides a means for data-driven controller design in generalized predictive controller (GPC) framework. Finally, the proposed control approach is illustrated with a verification study of a nonlinear continuously stirred tank reactor (CSTR) system. Copyright © 2011 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society

Data-driven predictive control for blast furnace ironmaking process

High performance control of blast furnace (BF) ironmaking process is a difficult problem due to the high temperature and hostile measurement conditions for measuring devices in the process. Previous research focused on developing of accurate predictive models for silicon content in hot metal ([SI]) while control of the whole process is seldom discussed. In the present work, a data-driven predictive control method based on subspace method is presented for the blast furnace ironmaking process. The algorithm is based on input–output data and easy to implement. Simulation results show the algorithm is effective for the control application. Finally, various practical issues concerning predictive control of blast furnace ironmaking process are also addressed, such as constraint handling, control objective and output set-point selection, adaptive strategy, etc.

Data-driven predictive control for solid oxide fuel cells

This paper is concerned with predictive control of solid oxide fuel cells (SOFC) based on a benchmark model commonly studied in the dynamic SOFC modeling/control literature. It has been shown in previous studies that control of SOFC is challenging owing to the slow response and tight operating constraints. In this paper, we apply a data-driven predictive control approach to solving the control problem of the SOFC system. The predictive control applied is completely data based. In addition, unlike other data-driven predictive control designs, the proposed approach can deal with systems without complete on-line measurement of all output variables. Simulation results have demonstrated the feasibility of the control application.