Robust control of seismic structures using independent modal-space techniques

Active robust structural controls have been utilized in the control of aerospace structures for many years but they have only been recently investigated in the context of control for civil engineering structures. The results of an investigation of the utilization of these methods on building-like structures are presented in this paper. The closed-loop systems take into account the limited available actuation force and are inherently insensitive to parameter variations and modeling uncertainties. Independent modal-space control (IMSC) is a structural control technique where the multi-input-multi-output configuration-space system is transformed into a set of uncoupled single-input-single-output modal-space systems. A modal controller is designed for each modal-space system and the set of modal controllers is transformed back into configuration-space. By combining IMSC with robust control techniques such as LQG/LTR or H _∞, a robust structural control design technique is proposed in this paper. Robust IMSC techniques are employed for control of seismic structures where a small number of actuators are used to control the first few modes of the structure. We have designed and implemented robust IMSC controllers on an experimental building-like structure. This structure utilizes torque motor driven active tendons as actuators and rests on a shaking table which is capable of providing one dimensional base excitation similar to earthquake ground motion. A three input-three output model of the structure, including the torque motor actuators, was developed using experimental data. The experimental structural identification technique, based on standard modal analysis methods, provides the mathematical model that describes the behavior of the structure. An H _∞ based IMSC controller has been designed and implemented on this structure using a dSPACE control development system. The results show that the performance of the system is satisfactory in the presence of unmodeled dynamics, parameter variations, and disturbance inputs.