Horizon specific expert fusion improves short term solar power forecasts
Horizon-specific Expert Fusion for Photovoltaic Power Forecasting
Artificial Intelligence
Summary
Forecasting how much solar power will be generated shortly ahead is tricky because sunlight varies regularly with the sun and unpredictably with weather. The authors developed a method that mixes different specialized models—including neural networks, historic examples, and weather predictions—using weights that depend on how far ahead the forecast is. This method was tested on public datasets and found to reduce forecasting errors compared to some strong individual models. However, the improvement depends on the dataset and timing.
What this means in practice
- •For electric grid operators: Improve short-term solar power forecasts by combining multiple expert models with weights tuned to different forecast horizons.
- •For solar power plant managers: Use a horizon-specific ensemble approach to reduce forecast errors and better plan operational decisions for photovoltaic generation.
Authors
Xu Yuqing, Zhou Liguo, Sun Ze, Yu Lei, Jiang Mingming
Abstract
Short-term photovoltaic power forecasting requires models to represent regular solar cycles and weather-driven fluctuations whose importance changes with the forecast horizon. This study develops a hierarchical ensemble that combines temporal neural models, historical analogs, state climatology, and gradient-boosted trees. Solar geometry and numerical weather forecasts describe the expected generation conditions, while horizon-specific convex weights combine complementary predictions. A separate calibration step uses available historical forecast errors to account for recent bias. The framework is evaluated on public PVDAQ data at 15--240-minute horizons and on three GEFCom2014 solar zones at hourly horizons up to four hours. On PVDAQ, the ensemble achieves a daylight capacity-normalized mean absolute error of 4.315%, reducing error by 4.11% relative to full-feature LightGBM and by 6.03% relative to fine-tuned Chronos-2 under identical calibration. Expert-removal experiments identify redundancy within the ensemble. Across three training seeds on GEFCom2014, learned fusion improves upon equal weighting but performs comparably to LightGBM. The results support horizon-specific combination as a useful forecasting strategy while showing that its advantage over strong individual models depends on the dataset and evaluation period.