Demand response-based central controller for energy management system using Fuzzy Double Q-Learning model

The rapid depletion of fossil reserves and their rising costs necessitate that consumers collaborate and use their resources judiciously. This paper proposes a demand response-based centralized control strategy that facilitates optimal resource management. A central controller (CC) is developed to overlook the buy/sell strategies based on load, generation status and utility retail prices. This community ecosystem consists of smart and regular houses, renewable energy sources (RES) and a community-wide energy storage system (ESS). The CC facilitates community energy brokering with the grid via these smart houses while still aligning with community requirements through the integration of RES and ESS. A novel CC leveraging Fuzzy Double Q-Learning method is introduced, and it is benchmarked against a Deterministic method and an established Fuzzy Single Q-Learning method. A substantial improvement in energy trading performance, buying and selling flexibility, and cost to community (C2C) is observed via this model–methodology combination. The comprehensive comparative analysis also showcases the significance of the cost reductions, the improved efficiencies, and the sustainable and economic approach toward the community ESS. The results indicate an overall improvement of 18.3% in C2C and a day saving of $5.34 when compared against Deterministic approach. Additionally, a 3.4% improvement in C2C is observed when compared against an established Fuzzy Single Q-Learning method.