Privacy Protection in Koopman Predictive Control

As the generation and utilization of data have become increasingly prevalent, the protection of personal privacy has become paramount. In the field of control systems, Koopman predictive control has emerged as an effective method for predicting and controlling systems using dynamic models. However, traditional Koopman predictive control methods pose privacy risks in terms of data usage and information sharing. To address this issue, privacy-preserving approaches for Koopman predictive control have been developed. In this paper, we propose a privacy-preserving Koopman predictive control method that employs privacy protection techniques to safeguard personal data. Specifically, we encrypt the measurements and control signals transmitted over the network channel, resulting in a simple and feasible data-driven encrypted predictive control algorithm. Then, the effectiveness of this method was verified through MATLAB simulation. Finally, an evaluation was made on the privacy level of the proposed algorithm.