An adaptive Home Energy Management system for prosumers in peer-to-peer trading networks with machine learning optimization

The proliferation of advanced metering systems has enabled decentralized energy management, allowing prosumers to optimize usage and trading. This study introduces a machine-learning enhanced HEMS framework operating in three stages: asset scheduling, bid optimization, and real-time adjustment. Results from a simulated community of four prosumers demonstrate a 30% reduction in grid dependency, a 20% increase in revenue, and an 18% decrease in CO 2 emissions. Interval-based uncertainty modeling further enhances robustness. This framework improves participation and economic returns in competitive peer-to-peer trading networks. • Machine learning improves peer-to-peer (P2P) trading forecasts. • Home energy assets are optimized across PV, EV, HVAC, and storage. • A three-stage HEMS model increases revenue and reduces grid dependency. • Interval analysis enhances system resilience to uncertainty. • Real-time adaptability demonstrated in a four-prosumer case study.