Stanford sky images and PV power generation dataset for solar forecasting related research and applications

SFML is the first fully local AI solar forecast for Home Assistant, powered by a local Attention Transformer. No external AI — such as ChatGPT, Gemini, or Grok — required. Runs entirely on your device for complete privacy.

list of papers, code, and other resources

Multi-time-horizon solar forecasting using recurrent neural network

PyTorch implementation of LSTM Neural Network for Multi-time-horizon solar forecasting

Short-Term Solar Forecasting Using LSTMs

Optimize the self consumption of your solar panels and measure your electricity consumption.

Official code for arXiv:2604.11807 - Physics-Informed State Space Models for Off-Grid Solar Forecasting

Python implementation of the Smart Persistence Model (SPM) for short-term photovoltaic power forecasting

An R-based solar power forecasting framework utilizing Support Vector Regression (SVR). Features include One-Class SVM outlier removal, diurnal cyclic encoding, and a robust CLI interface via Docker.

Official code: Physics-Informed Cross-Attention Networks for Solar Irradiance Forecasting with Dual Self+Cross Attention

Official code for arXiv:2604.11909 - Thermodynamic Liquid Manifold Networks for Autonomous Microgrids

Solar supply forecasting model training and MLflow deployment

Autonomous Sigen inverter controller for home solar and battery systems. Very lightweight (can run on RPi), making mode decisions every 5 minutes based on solar forecasts and live battery SOC — optimising between, self-consumption, grid export, and overnight TOU charging. Using Irish ESB Solar forecasts but portable to other providers.

Official code for arXiv:2604.13455 - Physics-Guided CNN-BiLSTM for Solar Irradiance Forecasting