Event-based predictive control triggered by input and output deadband conditions

This paper describes a predictive event-based control algorithm focused on practical issues. The Generalized Predictive Control scheme is used as predictive controller, and sensor and actuator deadbands are included in the design procedure to provide event-based capabilities. The first configuration adapts a control structure to deal with asynchronous measurements; the second approach is used for asynchronous process updates. The presented ideas, for an event-based control scheme, conserve all of the well-known advantages of the adaptive sampling techniques applied to process sampling and updating. The objective of this combination is to reduce the control effort while the control system precision is maintained at an acceptable level. The presented algorithm allows to obtain a direct tradeoff between performance and number of actuator events. The diurnal greenhouse temperature control problem is used as test bench to evaluate the different control algorithms configurations.