In this paper, an observer-based feedback/feedforward model predictive control (MPC) algorithm is developed for addressing the active vibration control (AVC) of lightly damped structures. For this purpose, the feedback control design process is formulated in the framework of disturbance rejection control (DRC) and a repetitive MPC is adapted to guarantee the robust asymptotic stability of the closed-loop system. To this end, a recursive least squares (RLS) algorithm is engaged for online estimation of the disturbance signal, and the estimated disturbance is feed-forwarded through the control channels. The mismatched disturbance is considered as a broadband energy bounded unknown signal independent of the control input, and the internal model principle is adjusted to account for its governing dynamics. For the sake of relieving the computational burden of online optimization in MPC scheme, within the broad prediction horizons, a set of orthonormal Laguerre functions is utilized. The controller design is carried out on a reduced-order model of the experimental system in the nominal frequency range of operation. Accordingly, the system model is constructed following the frequency-domain subspace system identification method. Comprehensive experimental analyses in both time-/frequencydomain are performed to investigate the robustness of the AVC system regarding the unmodeled dynamics, parametric uncertainties, and external noises. Additionally, the spillover effect of the actuation authorities and saturation of the active elements as two common issues of AVC systems are addressed.KEYWORDS disturbance estimation, model predictive control, orthonormal basis functions, piezoelectric material, vibration attenuation
| INTRODUCTIONModel predictive control (MPC) is an optimization-based method in which, the control inputs are obtained by solving a set of finite-horizon optimal control problems successively over the sampling instants. The optimal control formulation of the MPC explicitly takes into account model-based predictions of the controlled system trajectories. The obtained solutions of the optimal control problems are concatenated in a receding horizon framework, to construct the MPC feedback law. By the advances of embedded computer processors along with the enhanced optimization solvers, the MPC methods are appearing more and more in the emerging industrial applications. [1] In comparison with the other industrial control