摘要：AbstractA question that faces data-driven autonomous systems is verification that they will perform in a safe manner despite changes in the environment on which they act over time or incomplete knowledge of the system model. This work analyzes closed-loop stability of nonlinear systems under Lyapunov-based economic model predictive control (LEMPC) with data-driven models in the case where it is desirable to have the ability to detect when the data-driven model is or becomes insufficiently accurate for maintaining the closed-loop state in an expected region of state-space. Implications of the results for false sensor measurement cyberattacks seeking to impact the fidelity of models derived from model identification are discussed and illustrated through a chemical process example.
关键词：Keywordsmodel predictive controlanomaly responsechemical process controlnonlinear systemsempirical modelingcybersecurity
