Short-Term Reservoir Inflow Prediction Using Weather Forecast Data and ExtraTrees Regressor for Hydroelectric Power Plant Operations
Abstract
Reliable short-term inflow forecasting is critical for optimizing hydroelectric power plant (HPP) operations, where water availability is directly influenced by dynamic and often extreme weather patterns. Inaccurate inflow estimations can result in either uncontrolled water spills or reservoir levels dropping below operational thresholds, both leading to reduced energy efficiency and potential operational risks. This study presents a data-driven inflow prediction framework utilizing the ExtraTrees Regressor algorithm, trained on four years of combined historical hydrological and meteorological data, and validated using independent weather forecast datasets. The model achieves a coefficient of determination (R²) of up to 82%, demonstrating strong predictive accuracy. By incorporating forecasted weather parameters into the predictive model, the proposed approach enables proactive reservoir management and more effective turbine scheduling, significantly reducing the incidence of emergency spill events. Findings from this research highlight the advantages of transitioning from conventional intuition-based methods to systematic data-driven strategies, offering enhanced reliability, operational resilience, and efficiency in hydropower generation planning.