Track: Energy

Abstract

Renewable energy has gained immense attention due to its potential to reduce the world's dependency on fossil fuels and mitigate climate change. As the energy harnessing rate from renewable sources depends on some external parameters like solar radiation, wind speed, direction, and turbulence, the generation rate in such sources comes with a high level of fluctuations. Fluctuations in their output can increase operating costs for the electricity system and be quite challenging for utility companies to always maintain a proper balance between the generation and usage of electricity. To reduce the operation cost and increase the reliability of the system, robust forecasting models are used to predict the generation rate and energy demand. This comprehensive review paper offers a thorough examination of cutting-edge data-driven forecasting models utilized in forecasting renewable energy generation and demand. The paper organizes previous studies into five distinct groups based on prediction time frame: immediate, very short-term, short-term, medium-term, and long-term. It subsequently assesses the performance of various forecasting models, including three primary categories: time series, machine learning, and ensemble models, with respect to predicting energy demand and generation rates across different time frames, using standard performance evaluation metrics. The findings indicate that ensemble models employing neural networks and support vector machines demonstrate notably higher accuracy rates in predicting energy demand and generation rates in compared to the other models