Security of chemical process control systems against cyberattacks is critical due to the potential for injuries and loss of life when chemical process systems fail. A potential means by which process control systems may be attacked is through the manipulation of the measurements received by the controller. One approach for addressing this is to design controllers that make manipulating the measurements received by the controller in any meaningful fashion very difficult, making the controllers a less attractive target for a cyberattack of this type. In this work, we develop a model predictive control (MPC) implementation strategy that incorporates Lyapunov-based stability constraints and can allow several potential control laws to be available to apply to the process, one of which can be randomly selected at each sampling time, potentially making the response of the controller to a false state measurement more difficult to predict a priori. We investigate closed-loop stability and recursive feasibility of the resulting control design, and utilize a benchmark chemical process example to demonstrate the difference in the control actions computed by such a randomized MPC implementation strategy compared with those for the same process by the same MPC design utilized at every sampling time.
Controls and Control Theory | Information Security | Process Control and Systems
Durand, H., “State Measurement Spoofing Prevention through Model Predictive Control Design,” Proceedings of the 6th IFAC Conference on Nonlinear Model Predictive Control, 643-648, Madison, Wisconsin, 2018. DOI: 10.1016/j.ifacol.2018.11.034.