Acceleration feedback in model predictive control of electromechanical drive systems.

Abstract

This thesis discusses the use of acceleration measurements in predictive control schemes for electromechanical positioning systems. The proposed method is intended to improve the tracking error while maintaining the controllability and the robustness of the' controllers. The influence of acceleration feedback on controllers and observers of predictive type is evaluated, and the results are compared with those obtained by classical control schemes, without this type of feedback. Multiple approaches are subsequently compared, from the classical PID, Linear Quadratic Gaussian (LQG) control, to Dynamic Matrix Control (DMC) with state feedback and DMC with acceleration feedback (DMC/AF) as well. The simulations and the experiments are done using the framework of MATLAB/Simulink and dSPACE computer packages, on an electromechanical positioning system (EMPS) with friction wheel used at the Cologne Laboratory of Mechatronics (CLM). A controller design suite using GUI developed to facilitate industrial implementation has been used for rapid testing of diverse schemes. The observers for non-measured states were of Kalman filters type. (Abstract shortened by UMI.)