Short-term photovoltaic power forecasting using Artificial Neural Networks and an Analog Ensemble


Real power generation data from three solar farms are used.

A combination of ANN + AnEn outperforms each method run individually.

Probabilistic forecasts have high statistical scores using different verification measures.

Optimization parameters for the AnEn are investigated.

The scalability is tested using extreme scale (>150,000 cores) supercomputer simulations.


A methodology based on Artificial Neural Networks (ANN) and an Analog Ensemble (AnEn) is presented to generate 72 h deterministic and probabilistic forecasts of power generated by photovoltaic (PV) power plants using input from a numerical weather prediction model and computed astronomical variables. ANN and AnEn are used individually and in combination to generate forecasts for three solar power plants located in Italy. The computational scalability of the proposed solution is tested using synthetic data simulating 4450 PV power stations. The National Center for Atmospheric Research (NCAR) Yellowstone supercomputer is employed to test the parallel implementation of the proposed solution, ranging from one node (32 cores) to 4450 nodes (141,140 cores). Results show that a combined AnEn + ANN solution yields best results, and that the proposed solution is well suited for massive scale computation.


  • Solar power;
  • numerical weather prediction;
  • Artificial Neural Networks;
  • Uncertainty estimation;
  • Ensemble modeling;
  • Parallel computing

Be the first to comment

Leave a Reply

Your email address will not be published.