VRN3P: Variational Recurrent Neural Network Based NetLoad Prediction under High Solar Penetration

Summary

The increasing penetration of intermittent renewable energy sources like solar creates a challenge for maintaining a stable and cost-effective power supply due to forecasting uncertainties. This project developed VRN3P, a deep variational recurrent neural network-based framework for more accurate probabilistic day-ahead net-load forecasting, especially under high solar penetration. The project is completed, resulting in an open-source, validated tool demonstrating significant improvements in forecast performance, training time, and memory efficiency.

Key Facts

Source Document

https://example-government.gov/policy-document-link