(564g) Fault-Tolerant Model Predictive Control of Nonlinear Process Systems Using a Forecast-Triggered Communication Strategy

Model predictive control (MPC), also known as receding horizon control, refers to a class of optimization-based control algorithms that utilize an explicit process model to predict the future response of a plant. At each sampling time, a finite-horizon optimal control problem with a cost functional that captures the desired performance requirements is solved subject to state and control constraints, and a sequence of control actions over the optimization horizon is generated. The first part of the control inputs in the sequence is implemented on the plant, and the optimization problem is solved repeatedly at every sampling time. Originally developed to meet the specialized control needs of power plants and petroleum refineries, MPC technology can now be found in a wide variety of application areas including chemicals, food processing, automotive, and aerospace applications (e.g., see [1]). Motivated by the advantages of MPC, such as constraint handling capabilities, performance optimization, handling multi-variable interactions and ease of implementation, an extensive and growing body of research has been developed over the past few decades on the analysis, design and implementation of MPC, leading to a plethora of MPC formulations (e.g., see [2], [3] for some recent research directions and references in the field).

 Over the past few decades, and with the increasing demand for system performance and stability, fault-tolerance capabilities have become a critical component of modern day control system, especially for safety-critical systems, such as chemical processes, where malfunctions in the control actuators, measurement sensors and process equipment can lead to instabilities and safety hazards (e.g., see [4], [5] for some results and references on fault-tolerant control) and there are increasing calls to achieve zero-incident plant operations [6]. MPC, as a conventional feedback controller design methodology, is also faced with the challenges of dealing with device faults and handling the resulting deterioration in the closed-loop stability and performance requirements. While various methods have been investigated and developed for the design and implementation of fault-tolerant MPC for both linear and nonlinear processes (e.g., see [7]-[11]), the majority of existing methods at this point have been developed within the conventional feedback control setting where the sensor-controller communication links are assumed to be essentially dedicated links with flawless information transfer.

 With the advent of networked control systems, however, and the subsequent increased levels of integration of communication networks in the feedback loop, this conventional design setting needs to be re-examined to address the fundamental challenges introduced by the intrinsic limitations on the processing and transmission capabilities of the sensor-controller communication medium. Examples of efforts to address this problem within the MPC framework include the results in [12], [13] where resource-aware MPC formulations have been developed using techniques from event-based control to enforce closed-loop stability with reduced sensor-controller communication requirements. In these studies, however, the problem of incorporating fault-tolerance capabilities in the MPC design framework was not addressed.

 Motivated by these considerations, this work presents a methodology of the design of fault-tolerant MPC system for constrained nonlinear process systems, subject to model uncertainties, control actuator faults and sensor-controller communication constraints. The co-presence of faults, control and communication resource constraints creates a conflict in the control design objectives where, on the one hand, increased levels of sensor-controller communication may be required for robust fault handling, but, on the other hand, such levels may be either undesirable or unattainable due to the intrinsic limitations on the communication medium. To resolve this conflict, a resource-aware Lyapuonv-based MPC system that achieves the fault-tolerant stabilization objective with reduced sensor-controller communication is designed. In this approach, the control action is computed by solving on-line a finite-horizon optimal control problem based on an uncertain model of the plant subject to appropriate Lyapuonv-based stability constraints. The stability constraints are designed to ensure the desired closed-loop stability and performance properties in the presence of faults, and an explicit characterization of the state space region where fault-tolerant stabilization is guaranteed is obtained in terms of the fault size, the choice of the controller design parameters and the size of the plant-model mismatch. To keep sensor-controller communication to a minimum, a forecast-triggered communication strategy is used to determine when communication should be suspended or restored over the network. In this strategy, an update of the model state in the predictive controller using the sensor measurements at a given sampling time is triggered only when the Lyapunov function is forecasted to breach a certain threshold over the next sampling interval. The update-triggering threshold is derived using Lyapunov techniques and is explicitly parameterized in terms of the fault and a suitable fault accommodation parameter. Based on this characterization, fault accommodation strategies that guarantee closed-loop stability while simultaneously optimizing control and communication system resources are devised. Finally, the developed MPC formulation is illustrated using a chemical process example.

References:

[1] Qin, S.J. and Badgwell, T.A. “A survey of industrial model predictive control technology,” Control Engineering Practice, 11(7), 733–764, 2013.

[2] Rawlings, J.B. and Mayne, D.Q. “Model Predictive Control: Theory and Design,” Nob Hill Publishing, Madison, WI, 2009.

[3] Ellis, M., Liu, J., and Christofides, P.D. “Economic Model Predictive Control: Theory, Formulations and Chemical Process Applications,” Springer-Verlag, London, 2017.

[4] Blanke, M., Kinnaert, M., Lunze, J., and Staroswiecki, M. “Diagnosis and Fault-Tolerant Control,” Springer, Berlin-Heidelberg, 2003.

[5] Zhang, Y. and Jiang, J. “Bibliographical review on reconfigurable fault-tolerant control systems,” Annu. Rev. in Contr., 32, 229–252, 2008.

[6] Christofides, P.D., Davis, J., El-Farra, N.H., Clark, D., Harris, K., and Gipson, J. “Smart plant operations: Vision, progress and challenges,” AIChE J., 53(11), 2734–2741, 2007.

[7] Mhaskar, P. “Robust model predictive control design for fault-tolerant control of process systems,” Ind. Eng. & Chem. Res., 45, 8565–8574, 2006.

[8] Dong, J., Verhaegen, M., and Holweg, E. “Closed-loop subspace predictive control for fault-tolerant MPC design,” In Proceedings of 17th IFAC World Congress, 2008.

[9] Camacho, E.F., Alamo, T., and de la Pena, D.M. “Fault-tolerant model predictive control,” In Proceedings of IEEE Conference on Emerging Technologies and Factory Automation, 1–8. Bilbao, Spain, 2010.

[10] Lao, L., Ellis, M., and Christofides, P.D. “Proactive fault-tolerant model predictive control,” AIChE J., 59, 2810–2820. 2013.

[11] Knudsen, B.R. “Proactive actuator fault-tolerance in economic MPC for nonlinear process plants,” In Proceedings of 11th IFAC Symposium on Dynamics and Control of Process Systems, 1097–1102. Trondheim, Norway, 2016.

[12] Hu, Y. and El-Farra, N.H. “Quasi-decentralized output feedback model predictive control of networked process systems with forecast-triggered communication,” In Proceedings of American Control Conference, 2612–2617. Denver, CO, 2013.

[13] Hu, Y. and El-Farra, N.H. “Adaptive quasi-decentralized MPC of networked process systems,” In Distributed Model Predictive Control Made Easy, volume 69, 209–223. J. M. Maestre and R. R. Negenborn (Eds.), Springer, Netherlands, 2014.