Model-based predictive control: Tracking performance, robustness, and its application to a 3-DOF motion simulator.

摘要

The original purpose of this dissertation is to develop a set of systematic methods to measure tracking performance and robustness of a model-based predictive control system in the frequency domain, to derive a performance sensitivity representation (the indicator of performance robustness) for a special kind of multi-input/multi-output (MIMO) model-based predictive closed-loop control system, in which the plant has equal number of inputs and outputs, and to be the first to report a successful application of a model-based predictive controller to a 3 degree-of-freedom electro-hydraulic automobile seat simulator.;A frequency domain interpretation of the model-based predictive control system was well developed in this research for performance and robustness analysis. Taking the motion simulator as a design example, the indicators of robust performance and robust stability were established for the SISO model, and then generalized to the MIMO case.;Although model-based predictive controllers are generally designed in the discrete time domain, a model-based predictive controller that was constructed in the continuous time domain was also studied in this research. The robustness analysis in frequency domain was first conducted for this kind of continuous model-based predictive closed-loop control system.;The discrete MIMO model-based predictive controller with the selected design parameters through trade-off analysis was implemented on the seat simulator. It was experimentally proven that these kinds of feedback/feedforward controllers produce good tracking and robust properties.;In this study, the procedures for predictive control design in the discrete time domain for both single-input/single-output (SISO) and multi-input/multi-output (MIMO) models were performed. Much attention has been paid to a thorough analysis of the influence of all design parameters of model-based predictive controllers on tracking performance and robustness of the resultant closed-loop system. The controllers were designed diligently to balance tracking performance, control energy and robustness.;Finally, the derivation of a robust performance indicator in the frequency domain for more general multi-input/multi-output models (whose inputs do not equal to their outputs), and the robustness analysis of a constrained and nonlinear model-based predictive controller are highly recommended for future work.