Elevator Speed Tracking Using Event-Triggered Model Predictive Control with Learning-Based Initialisation

This paper is concerned with the problem of elevator speed tracking. To reduce the computational complexity of the standard model predictive control (MPC), we propose an event-triggered MPC method that guarantees control effectiveness. This method includes two control stages: initialisation and online optimisation. During the initialisation stage, a supervised learning technique is employed to approximate the MPC using sample data. The online optimisation stage involves controlling the elevator system to track an ideal speed curve with a designed event-triggering mechanism. The proposed method is evaluated against the standard MPC in the simulation by tracking various speed curves. The results demonstrate that the proposed method significantly reduces computational time while preserving tracking accuracy, making it more suitable for real-world elevator systems.