Modified predictive control of structures with direct output feedback

Abstract

The algorithm of modified predictive control (MPC) is derived with the partial-state concept of direct output feedback (DOF) to reduce the number of sensors for real implementation. The feasibility of using modified predictive control with direct output feedback (MPCwDOF) is verified through numerical and experimental study. According to the MPCwDOF algorithm, the online control forces are simply generated from the actual output measurements that are multiplied by a prescribed constant output feedback gain matrix. An off-line numerical method is introduced to find the feedback gain systematically and efficiently. Numerical examples of two control systems with respect to single-controller and multiple-controller systems are illustrated for validating the feasibility of using the MPCwDOF algorithm and the application of the decentralized control strategy to the MPCwDOF algorithm. An experiment of a large-scale 5-story structural model with an MPCwDOF-controlled active mass damper on the roof is also performed. Results indicate that the controlled structures achieve good performance under environmental excitations. Simple online calculations and a small number of sensors make the proposed control algorithm more favorable to real implementation. Copyright © 2010 John Wiley & Sons, Ltd.