Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
March 4, 2026SustainabilityOpen Access

Improving Photovoltaic Power Forecasting Accuracy by Integrating Aerosol Optical Features: A Dual-Channel Deep Learning Approach

View Full Paper
Ask AI
Bookmark
Share

Authors

TYTing YangBCButian ChenQCQi Cheng

Discussion

Loading...

Member takes

Overview

This research integrates aerosol optical features with deep learning to enhance power forecasting accuracy in varying atmospheric conditions.

Key Points

  • The aim is to enhance short-term photovoltaic power prediction by integrating aerosol optical features.
  • Developed a dual-channel encoder–decoder network using BiLSTM and iTransformer.
  • Constructed a high-dimensional aerosol optical feature set from OPAC database.
  • Optimized features with minimum redundancy maximum relevance (mRMR) algorithm.
  • Categorized prediction scenarios into polluted and clean regimes through K-means clustering.
  • Validated method with data from a PV station and Copernicus Atmosphere Monitoring Service.
  • Achieved approximately 29.83% reduction in mean absolute error (MAE) on polluted days.
  • Achieved approximately 15.22% reduction in MAE on clean days.
  • Highlighted distinct physical mechanisms driving predictions related to extinction and scattering.

Cite This Study

Yang et al. (2026) studied this question.

synapsesocial.com/papers/69a7cdaed48f933b5eeda3c5https://doi.org/10.3390/su18052403
View Full Paper
Ask AI
Bookmark
Share