Operational Value of Multi-Horizon Load Forecasts for Energy Management in a Hydrogen-Enabled Community Microgrid

Abstract

Hydrogen-enabled community microgrids can improve renewable energy use and local resilience, but their operation is complicated by uncertain residential demand, variable renewable generation, dynamic electricity prices, and coupled battery and hydrogen storage. This paper evaluates the operational value of multi-horizon community load forecasts when incorporated into a previously developed proximal policy optimization (PPO) energy management system for a 1,000-household residential microgrid in Rockhampton, Australia. Forecast accuracy is mixed. The 1-hour horizon achieves an RMSE of 239.32 kW and an R-squared value of 0.201, while the 6-hour and 12-hour horizons produce negative R-squared values. The 24-hour forecast achieves an RMSE of 249.79 kW, a MAPE of 62.52 percent, and an R-squared value of 0.126. In the reported single-seed PPO experiment, the forecast-enriched controller achieves a final reward 8.3 percent higher than the non-predictive controller. Annual savings increase from AUD 2,439.86 to AUD 2,765.83, representing an additional AUD 325.97 or 13.4 percent relative to the non-predictive savings. Grid imports decrease to 58,147.49 kWh. These results provide proof-of-concept evidence rather than multi-seed validation.