An AI-Based Decision-Support Pipeline for Day-Ahead Photovoltaic Forecasting

2026-08-03

AI Summary

This study addresses the challenge of short and incomplete historical data at newly deployed photovoltaic sites, which severely limits the performance of conventional day-ahead forecasting models and consequently impacts charging and energy storage scheduling. To overcome this limitation, the authors propose a deployment-oriented environmental AI forecasting pipeline that incorporates timestamp calibration, leakage-free solar geometry and clear-sky index feature engineering, and integration of short-term meteorological context. Furthermore, they design a physics-informed stacked ensemble strategy validated via rolling-origin evaluation. The approach significantly enhances prediction robustness under data scarcity. Experimental results demonstrate that, compared to the smart persistence benchmark, the proposed method reduces daytime normalized RMSE by 32% and 9% under random daily-block and rolling-origin evaluation protocols, respectively. It also achieves approximately 6.5% lower daytime RMSE than the best single-model baseline, highlighting the critical influence of evaluation protocols and deployment scenarios on method effectiveness.

Abstract

Reliable photovoltaic (PV) forecasts are needed for low-carbon energy systems, but newly deployed sites often have short, imperfect records. This makes standard day-ahead forecasting difficult: persistence and physical baselines can be sensitive to calibration and timestamp alignment, while single machine-learning models may capture only one structure in the data and overstate skill under non-temporal validation. We study this problem at a United Kingdom charging-station site, where PV forecast errors affect charging availability, storage scheduling, and downstream control. Using measured inverter output and publicly available meteorological inputs, we develop a deployment-oriented environmental-AI pipeline for day-ahead hourly PV forecasting. The pipeline corrects timestamp conventions, constructs leakage-safe solar-geometry and clearness-index features, adds short-term atmospheric context, and combines complementary predictors through validation-learned stacking. Against smart persistence, a clear-sky baseline that adjusts recent PV output using expected clear-sky irradiance, the best ensemble reduces daylight normalised RMSE by about 32% under random day-blocked evaluation and 9% under the stricter rolling-origin protocol. It also reduces daylight RMSE relative to the strongest individual machine-learning baseline by 6.6% and 6.4%, respectively. The results show that physics-aware stacking can support PV forecasts from limited site data, but its value depends on model class, evaluation protocol, and deployment context.

Problem

• photovoltaic forecasting
• day-ahead prediction
• limited historical data
• forecasting uncertainty
• renewable energy integration

Innovation

• photovoltaic forecasting
• environmental AI
• leakage-safe features
• validation-learned stacking
• timestamp alignment