(196g) Cyberattack-Resilient Data-Gathering Policies Under Lyapunov-Based Economic Model Predictive Control for Physics-Based Model-Building

Data-gathering maneuvers by controllers have been considered in the context of, for example, dual model predictive control [4]. Attempts to use data in the identification of physics-based models have been pursued in, for example, [5,6]. However, it remains challenging to identify physics-based models from data. In [2,3], we attempted to use a controller to perform data-gathering maneuvers to identify physics-based models. However, an open question remains how to tell which data to gather to ensure that it identifies a physics-based model. On a different note, data has potential to be compromised if sensors are compromised. One of the main contributions of our recent work in [1] is that it elucidates how to detect attacks on an advanced controller known as Lyapunov-based economic model predictive control (LEMPC) [7] when sensor attacks,actuator attacks, or both can occur.

In this work, we consider two questions: 1) how to identify what data should be gathered for developing physics-based models and 2) how to extend our prior concepts on data-gathering to handling whether the data is correct or incorrect (both before the model is re-identified through detection and through analyzing the impacts of it being incorrect when the model is re-identified). We recall from [2,3] how LEMPC can be utilized for online data collection once desired data is identified. We then discuss methods for attempting to identify that data, including making use of concepts from Automated Theorem Proving. Specifically, Automated Theorem Proving (ATP) is a method that has seen applications in mathematical fields to automatically prove conjectures within a logical framework. This framework is formed from logical statements known as axioms, which are considered to be always true. In order to utilize the principles of ATP to aid in model validation, a method is proposed to attempt to verify whether a model is correct according to whether it meets certain axioms that should be satisfied by the data. We analyze this concept initially with a level in a tank process for a pre-developed axiom specifying that a correct model for the process should meet a certain requirement related to increases and decreases in height. We then explore how axioms might be automatically developed. After methods for identifying desired data are identified, we discuss how to use LEMPC for driving the closed-loop state toward that data as in [2,3], but add additional discussion regarding checking for cyberattacks during the data-gathering phase according to methods like those developed in [1,8]. We then also discuss how un-detected attacks during the data-gathering stage could impact model re-identification, considering a control-theoretic viewpoint in an LEMPC context integrated with the various detection strategies.

[1] Oyama, H., Messina, D., Durand, H., Rangan, K., “Lyapunov-Based Economic Model Predictive Control for Detecting and Handling Actuator and Simultaneous Sensor/Actuator Cyberattacks on Process Control Systems,” Frontiers in Chemical Engineering, 2022; 810129.

[2] Oyama, H., Durand, H., “Lyapunov-Based Economic Model Predictive Control for Online Model Discrimination,” Computers Chemical Engineering, 2022; 107769.

[3] Oyama, H., Leonard, A. F., Rahman, M., Gjonaj, G.,Williamson, M., and Durand, H., “Online Process Physics Tests via Lyapunov-based Economic Model
Predictive Control and Simulation-Based Testing of Image-Based Process Control,” American Control Conference, paper 1187, 2021.

[4] Heirung, T. A. N., Ydstie, B. E., & Foss, B. (2012). Towards dual MPC. IFAC Proceedings Volumes, 45(17), 502-507.

[5] Brunton, S. L., Proctor, J. L., & Kutz, J. N. (2016). Discovering governing equations from data by sparse identification of nonlinear dynamical systems. Proceedings of the national academy of sciences, 113(15), 3932-3937.

[6] Schmidt, M., & Lipson, H. (2009). Distilling free-form natural laws from experimental data. Science, 324(5923), 81-85.

[7] Heidarinejad, M., Liu, J., & Christofides, P. D. (2012). Economic model predictive control of nonlinear process systems using Lyapunov techniques. AIChE Journal, 58(3), 855-870.

[8] Oyama, H., & Durand, H. (2020). Integrated cyberattack detection and resilient control strategies using Lyapunov‐based economic model predictive control. AIChE Journal, 66(12), e17084.