Experimental verification of multirate and Model Predictive Control for discrete-valued control systems

Abstract

A plant with discrete-valued control is considered in this study. In discrete-valued control systems, the control input resolution, which is determined by the minimum value of the amplitude of the discrete-valued input and period, directly affects the control performance. If insufficiently short periods are specified, the control performance decreases due to the poor resolution of the discrete-valued input. To overcome such decrease, multirate control, which employs individual periods for output measurement and control input switching, was adopted in this study. We analyzed a decrease in the discrete-valued control performance caused by the poor control input resolution in pneumatic isolation table control, and numerical simulations and experiments showed that the use of multirate control is effective despite a long output period. The multirate control input was determined based on the Model Predictive Control, and a Kalman filter was employed in the experiments to reduce sensor noise for the pressure sensors.

Keywords

• discrete actuators
• discrete-valued control
• multirate control
• model predictive control
• quantized signals
• pneumatic isolation table