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

📄 arXiv: 2607.08079v1 📥 PDF

作者: Hang Fan, Weican Liu, Ying Lu, Dunnan Liu, Long Cheng, Wei Wei

分类: cs.AI

发布日期: 2026-07-09

🔗 代码/项目: GITHUB

💡 一句话要点

提出PARA-PV框架以解决光伏发电预测中的物理约束问题

🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 光伏发电 预测模型 物理约束 时间序列分析 机器学习 可再生能源 数据检索 智能电网

📋 核心要点

  1. 光伏发电预测面临天气变化和物理约束等多重挑战，现有方法难以有效整合这些因素。
  2. PARA-PV框架通过物理知识检索增强学习，结合历史数据和时间序列模型，提升预测准确性。
  3. 实验结果表明，PARA-PV在光伏发电预测中显著提高了准确性，尤其在关键状态下表现优异。

📝 摘要（中文）

准确的光伏发电预测对于可靠的电网调度和可再生能源的整合至关重要，但由于天气变化、昼夜交替、状态依赖动态和严格的物理约束，预测仍然面临挑战。本文提出了PARA-PV，一个物理知识嵌入的检索增强框架，首先将多变量光伏观测编码为补丁级表示，然后通过物理知识检索增强学习器检索与当前时间窗口一致的历史补丁和类比轨迹，生成物理基础的预测。接着，通过轻量级残差适配器将基础预测与冻结的Chronos时间序列基础模型进行校准，以适应光伏特定动态。最后，物理约束损失函数对样本进行分区和自适应重加权，以防止主流状态抑制关键状态的学习。

🔬 方法详解

问题定义：本文旨在解决光伏发电预测中的物理约束和天气变化带来的不确定性，现有方法未能有效整合这些因素，导致预测准确性不足。

核心思路：PARA-PV框架通过物理知识的嵌入和历史数据的检索，结合时间序列模型，提供一个物理基础的预测，同时适应光伏特定的动态变化。

技术框架：该框架包括多个模块：首先编码多变量光伏观测为补丁级表示；然后通过物理知识检索增强学习器进行历史数据检索；接着使用轻量级残差适配器校准基础预测；最后应用物理约束损失函数进行样本分区和重加权。

关键创新：PARA-PV的创新在于将物理知识与历史数据检索相结合，形成一个物理基础的预测框架，克服了传统方法的局限性。

关键设计：框架中使用的损失函数根据样本的不同状态进行自适应重加权，确保在关键状态下的学习不被主流状态抑制，同时采用门控均值偏移和尺度校正来调整预测。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果显示，PARA-PV在光伏发电预测中相较于基线模型提高了预测准确性，尤其在高峰和夜间状态下，表现出更强的鲁棒性和适应性，具体提升幅度达到15%以上。

🎯 应用场景

该研究在可再生能源管理、智能电网调度和气候变化应对等领域具有广泛的应用潜力。通过提高光伏发电预测的准确性，能够更好地整合可再生能源，促进可持续发展。未来，PARA-PV框架还可以扩展到其他类型的可再生能源预测中。

📄 摘要（原文）

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.