Experimental Investigation of Model Predictive Control for Thermal Energy Storage System Using Artificial Intelligence

A model predictive control (MPC) strategy was developed using artificial intelligence (AI) and investigated using an experimental setup. The experimental system for the cooling operation includes a chiller, thermal energy storage (TES), heat exchangers, and variable-speed pumps. The air-conditioning space was replaced with a water tank, and the cooling load was assigned by an electric immersion heater. Control variables included the flow rates of pumps in the water loop for the primary side. In the MPC framework, artificial neural networks (ANNs) were used as surrogate models, and a metaheuristics optimization solver was employed to minimize the total operating costs. The developed AI-based MPC strategy could save operating costs by 9.7–22.5% compared to the rule-based control strategies that prioritize TES operation.