PARA-PV: Physics-Aware Retrieval-Augmented PV Prediction Based on Frozen Foundation Model and Distribution Shift Correction

2026-07-08

AI Summary

This work addresses the challenges of distribution shift and physical inconsistency in photovoltaic (PV) power forecasting caused by abrupt weather changes, diurnal cycles, and operational state transitions. To tackle these issues, the authors propose PARA-PV, a novel framework that integrates a physics-aware retrieval-augmented mechanism, a frozen time-series foundation model (Chronos), and a dynamic distribution shift correction module. The approach generates physically consistent initial forecasts by retrieving relevant historical segments, refines predictions through lightweight residual adapters, and enhances learning under critical operating conditions—such as peak generation, ramping events, and nighttime—via a gated mean-scale correction and a piecewise adaptive loss function. Experimental results demonstrate that PARA-PV significantly improves prediction accuracy and effectively mitigates distribution shifts in complex environmental scenarios.

Abstract

Accurate photovoltaic (PV) power forecasting is essential for reliable grid dispatch and renewable energy integration, yet it remains challenging because PV generation is jointly shaped by weather variability, day-night transitions, regime-dependent dynamics, and strict physical constraints. We propose PARA-PV, a Physics-Aware Retrieval-Augmented framework that embeds physical knowledge throughout the forecasting process. The framework first encodes multivariate PV observations into patch-level representations and, through a physics-aware retrieval-augmented learner, retrieves historical patches and analog trajectories that are consistent with the current window in temporal shape, power level, PV operating state, and intra-day period; this yields a physically grounded base forecast. To supplement local memory with broader temporal knowledge, the base forecast is then calibrated against a frozen Chronos time-series foundation-model prior through a lightweight residual adapter, so that general temporal regularities are adapted to PV-specific dynamics without overriding the physically grounded prediction. Because residual conditional distribution shifts persist when weather and diurnal regimes change, a physics-aware distribution shift correction module subsequently adjusts the preliminary forecast using power, weather, timestamp, and day/night conditions, applying gated mean-shift and scale corrections selectively. Finally, a physics-constrained loss function partitions the samples into peak, ramping, night-time, and regular regimes and adaptively reweights their error contributions, preventing the dominant regular regime from suppressing learning of operationally critical states. Our code is available at https://github.com/weican1103/PARA-PV.

Problem

• photovoltaic power forecasting
• weather variability
• physical constraints
• distribution shift
• diurnal transitions

Innovation

• Physics-Aware Forecasting
• Retrieval-Augmented Prediction
• Distribution Shift Correction
• Frozen Foundation Model
• Physics-Constrained Loss