Explicit model predictive control of split-type air conditioning system

Domestic air conditioners are a major source of energy consumption. In this study, utilizing real-time data from a public domain, a cascaded hardware in loop approach to the control of room temperature is considered. An inner loop to control the supply air temperature by adjusting the electronic expansion valve using a second-order plus delay time model is proposed. The room temperature control is considered the outer loop. A simplified lumped parameter representation of the thermal dynamics of the building is modelled in MATLAB using ordinary differential equations. A constrained multi parametric model predictive controller (mpMPC) is designed for both the control loops. The constraints include safety limits on the superheat and manipulation rates for the inner loop and a rate constraint on the reference signal in the outer loop. Model uncertainties like ambient temperature and thermal load variations (representing an office space) are considered for hardware in the loop testing of the proposed strategy. From performance analysis, using power spent and thermal comfort quantization, it is observed that the mpMPC scheme outperforms traditional control strategies.