Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cloud-Edge Collaborative Large Models for Robust Photovoltaic Power Forecasting

Nan Qiao, Shuning Wang, Sijing Duan, Wenpeng Cui, Yuzhe Chen, Qingchen Yang, Xingyuan Hua, Ju Ren

Abstract

Photovoltaic (PV) power forecasting in edge-enabled grids requires balancing forecasting accuracy, robustness under weather-driven distribution shifts, and strict latency constraints. Existing models work well under normal conditions but often struggle with rare ramp events and unexpected weather changes. Relying solely on cloud-based large models often leads to significant communication delays, which can hinder timely and efficient forecasting in practical grid environments. To address these issues, we propose a condition-adaptive cloud-edge collaborative framework *CAPE* for PV forecasting. *CAPE* consists of three main modules: a site-specific expert model for routine predictions, a lightweight edge-side model for enhanced local inference, and a cloud-based large retrieval model that provides relevant historical cases when needed. These modules are coordinated by a screening module that evaluates uncertainty, out-of-distribution risk, weather mutations, and model disagreement. Furthermore, we employ a Lyapunov-guided routing strategy to dynamically determine when to escalate inference to more powerful models under long-term system constraints. The final forecast is produced through adaptive fusion of the selected model outputs. Experiments on two real-world PV datasets demonstrate that *CAPE* achieves superior performance in terms of forecasting accuracy, robustness, routing quality, and system efficiency.

Cloud-Edge Collaborative Large Models for Robust Photovoltaic Power Forecasting

Abstract

Photovoltaic (PV) power forecasting in edge-enabled grids requires balancing forecasting accuracy, robustness under weather-driven distribution shifts, and strict latency constraints. Existing models work well under normal conditions but often struggle with rare ramp events and unexpected weather changes. Relying solely on cloud-based large models often leads to significant communication delays, which can hinder timely and efficient forecasting in practical grid environments. To address these issues, we propose a condition-adaptive cloud-edge collaborative framework *CAPE* for PV forecasting. *CAPE* consists of three main modules: a site-specific expert model for routine predictions, a lightweight edge-side model for enhanced local inference, and a cloud-based large retrieval model that provides relevant historical cases when needed. These modules are coordinated by a screening module that evaluates uncertainty, out-of-distribution risk, weather mutations, and model disagreement. Furthermore, we employ a Lyapunov-guided routing strategy to dynamically determine when to escalate inference to more powerful models under long-term system constraints. The final forecast is produced through adaptive fusion of the selected model outputs. Experiments on two real-world PV datasets demonstrate that *CAPE* achieves superior performance in terms of forecasting accuracy, robustness, routing quality, and system efficiency.
Paper Structure (35 sections, 11 theorems, 138 equations, 8 figures, 2 tables, 3 algorithms)

This paper contains 35 sections, 11 theorems, 138 equations, 8 figures, 2 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. PV Power Forecasting Methods.
  4. Large Models for Time-Series Forecasting.
  5. Cloud-Edge Collaborative Inference.
  6. System Model and Problem Formulation
  7. Multi-Site Cloud-Edge Forecasting System
  8. Mean-Field Cloud Delay and Resource-Consumption
  9. Long-Term Stochastic Optimization Formulation
  10. Virtual-Queue for Long-Term Constraints
  11. Design of CAPE
  12. Local Screening and Routing-Score Computation
  13. Lyapunov-Guided Routing under Mean-Field Cloud Congestion
  14. Fusion of Selected Predictors with Online Updates
  15. Long-Term Performance Guarantee
  16. ...and 20 more sections

Key Result

Proposition 3.1

Denote $\xi_{i,t}$ as a latent weather-site regime variable. If $\mathbf{y}_{i,t}^{*} \perp\perp \mathcal{S}_{i,t} \mid \bigl( \mathbf{X}_{i,t}^{\mathrm{loc}}, \xi_{i,t} \bigr),$ then the Bayes-optimal cloud-assisted predictor admits the decomposition $p( \mathbf{y}_{i,t}^{*} \mid \mathbf{X}_{i,t}^{

Figures (8)

  • Figure 1: System model of the proposed cloud-edge PV forecasting.
  • Figure 2: Overview of CAPE framework.
  • Figure 3: Router action traces at three routing-score levels, where higher scores lead to more frequent escalation to larger models.
  • Figure 4: Effect of the Lyapunov trade-off parameter $V$ on AUROC under different edge counts $N$.
  • Figure 5: Effect of the Lyapunov trade-off parameter $V$ on $DG$ under different edge counts $N$.
  • ...and 3 more figures

Theorems & Definitions (29)

  • Proposition 3.1
  • Theorem 4.1
  • proof
  • Proposition 4.1
  • proof
  • Proposition 4.2: Retrieval-quality dominated loss gap
  • proof
  • Lemma 4.1
  • proof
  • Lemma 4.2
  • ...and 19 more