Robustness of Deep Learning Models for PV Power Forecasting under NWP Forecast Errors: A Spatiotemporal and Physically Interpretable Analysis

2026-07-14

AI Summary

This study addresses the limitations of existing evaluations that overlook the temporal correlation, state dependence, and physical coupling inherent in numerical weather prediction (NWP) errors. To this end, it proposes a simulation-based, physics-constrained robustness evaluation framework that incorporates clear-sky modulated heteroscedastic perturbations and an Erbs-based solar irradiance reconstruction mechanism to generate dynamic, heteroscedastic, and radiometrically consistent NWP perturbations. By isolating input uncertainty propagation through synthetic photovoltaic power generation, the framework systematically benchmarks models including PatchTST, GRU, N-HiTS, and LightGBM. Results demonstrate that sequence-based models outperform tabular models under moderate to high perturbation levels, exhibiting superior noise filtering and temporal resilience. Interpretability analysis further reveals a shift in feature reliance—from future forecasts toward historical observations and physical priors—as perturbations intensify. Combined with Pareto analysis, these insights offer principled guidance for model selection across trade-offs among accuracy, robustness, and latency.

Abstract

Engineering use of AI forecasting models requires not only high nominal accuracy but also predictable behavior under uncertain inputs. In photovoltaic (PV) forecasting, this requirement is especially challenging because numerical weather prediction (NWP) errors are temporally correlated, state dependent, and physically coupled across variables. Existing evaluations, however, often rely on perfect forecast assumptions or simplistic perturbations that do not reflect these characteristics. This study presents a physically constrained robustness evaluation framework based on simulation, using virtual PV power as a controlled response variable to isolate the propagation of input uncertainty from confounders at the plant level. Six representative machine learning and deep sequence models, including PatchTST, GRU, N-HITS, and LightGBM, are evaluated under dynamic NWP perturbations with heteroscedasticity modulated by clear-sky conditions and Erbs reconstruction that preserves radiation consistency. The results show that sequence models provide stronger noise filtering and temporal resilience than a strong tabular baseline under medium to high disturbance regimes. SHapley Additive exPlanations (SHAP) and Integrated Gradients (IG) further support a feature reallocation tendency at the case level, in which predictive reliance shifts from corrupted future forecasts toward more stable historical observations and deterministic physical priors. A Pareto analysis of accuracy under clean conditions, robustness, and computational latency then translates these findings into engineering implications for robustness assessment and model selection under forecast uncertainty.

Problem

• PV power forecasting
• NWP forecast errors
• model robustness
• input uncertainty
• spatiotemporal correlation

Innovation

• physically constrained robustness
• spatiotemporal NWP perturbations
• virtual PV power simulation
• feature reallocation under uncertainty
• Pareto-aware model selection