Multiparametric stochastic model predictive control of indoor air temperature and humidity in buildings

Abstract

This paper develops the multiparametric (mp) programming version of the Scenario-Based Stochastic Model Predictive Control (SB-SMPC) method to improve the control performance of MPC under disturbance. The developed mpSB-SMPC controller calculates the control law expressions offline, thereby enabling its deployment through a chip or low-cost hardware instead of a dedicated computer, as in the case of the online SB-SMPC controller. The mpSB-SMPC controller is deployed to a typical residential house system in Delhi, which consists of air conditioning (AC), heater, and air handling unit (AHU) ventilation systems. The controller aims to control the temperature, relative humidity, and occupant comfort indoors by optimally manipulating the AC, heater, and AHU ventilation actuators. Closed-loop operation under the developed mpSB-SMPC controller for different seasons shows improved tracking performance by 2.66% and 0.7% for indoor temperature and relative humidity, respectively, compared to the standard, deterministic case of mpMPC. Moreover, the operation under the developed controller shows significant improvement in peak occupant discomfort reduction by 3.9% for the whole year compared to the mpMPC base case.