Smart Agent-Based Direct Load Control of Air Conditioner Populations in Demand Side Management

The rapid fluctuations of renewable resources, such as wind and solar, pose challenges to the reliability of the grid and make it difficult to integrate them into existing power systems. To address this uncertainty and variability in generation, direct load control is a promising method for managing the demand side. Thermostatically controlled appliances, such as air conditioners, can play a significant role in this process. However, to implement direct load control, the system operator needs a reliable estimate of the load's magnitude and the degree to which it can be shifted. This paper introduces a smart-agent architecture that utilizes a mathematical model of the aggregated power consumption of the load to forecast its behavior, even in the presence of changes imposed by the controller. To evaluate the system's performance, a numerical simulator was developed. It was observed that the system can adapt to changes in the system, retrain itself, and develop a new mathematical model to improve its prediction of the aggregated power consumption.