Semi-active control and optimal estimation of dynamic characteristics of structures by digital adaptive filters

A semi-active multivariable adaptive controller is designed by using a filtered-x NLMS optimizer, for a framed structure under an seismic disturbance that is modeled by colored noise, output from an innovations filter. There is no need for actuator and large power supply, but just some valves and a battery-size power supply to control the amount of flow of a hydraulic fluid through a by-pass variable-orifice channel in an energy dissipating system. It consists of a pistrn attached to A-shaped wind bracing of the building and a cylinder attached lo upper floor. It is assumed there is just a rough finite element model of the structure, so the augmented dynamic system comprised of structure/energy dissipating system/earthquake is estimated by use of AR and ARX models and weighted RLS optimizer. In addition, it is assumed there is no enough structural sensor, so we implement an Extended Kalman Filter to estimate state vector of the augmented system, and to reject the measurement noise. The state estimates are also used in the prediction phase of seismic signal, in order to search for optimal viscous damper coefficients that act as calibration measures for corresponding orifice size of the by-pass channels of damping systems. The simulation results show that the controllability, offered by this method is not so large as an active counterpart, energized by conventional actuators, but due to its low-power requirement, it is comparable and sometimes superior