Evaluation of Machine Learning Techniques for Green Energy Prediction

TL;DR

This paper compares various machine learning methods like Bayesian inference, neural networks, and SVMs to predict green energy output from weather data, aiming to improve accuracy and data recovery.

Contribution

It provides a comprehensive evaluation of multiple ML techniques for green energy prediction and data recovery from weather datasets.

Findings

Neural networks outperform other models in prediction accuracy.

Clustering techniques help identify correlations among weather parameters.

Support Vector Machines effectively recover missing weather data.

Abstract

We evaluate the following Machine Learning techniques for Green Energy (Wind, Solar) Prediction: Bayesian Inference, Neural Networks, Support Vector Machines, Clustering techniques (PCA). Our objective is to predict green energy using weather forecasts, predict deviations from forecast green energy, find correlation amongst different weather parameters and green energy availability, recover lost or missing energy (/ weather) data. We use historical weather data and weather forecasts for the same.