Using Receding Horizon Minimization of Output Errors in Time Domain for Model Predictive Control of Multivariable Paper Drying Systems

Abstract

The paper proposes a model predictive control method for MIMO system, in which the difference system equations are used for output prediction. The proposed control method is based on piecewise minimizing of output tracking errors in time domain. The structure of proposed controller is similar to that of conventional GPC, except that its system output prediction and minimization of tracking error will be carried out directly in time domain instead of frequency domain. Hence the issue of Diophantine equation can be avoided but the tracking control performance is acceptable and stability is still guaranteed. Moreover, the decoupling and disturbance attenuation behavior of the proposed controller is better than the conventional GPC. To demonstrate these assertions, the paper presents an application of proposed method to control the multivariable paper drying system and then compares the resulted system control performance with the one using the output feedback controller in comparison with the feedforward controller for conventional GPC.