Predictive Modelling of Renewable Energy Growth with Machine Learning and Evidence Markov Chain

Abstract

This paper presents an advanced computational framework that integrates advanced machine learning curve fitting algorithms with an enhanced Markov Chain model to simulate and forecast the growth of the renewable energy sector in the United States. The proposed model incorporates Dempster-Shafer Evidence Theory to manage the inherent uncertainties in renewable energy production, producing refined probability distributions across diverse energy sources, such as biomass, wind, solar, and hydroelectric power. These distributions, used within a Monte Carlo simulation, enable robust and accurate predictions of future renewable energy trends. The application of machine learning plays a pivotal role in optimizing the model's performance, allowing for precise adjustments to the probabilistic structure and improving prediction accuracy. The model’s efficacy was demonstrated through accurate forecasts for the years 2017 and 2018. This study offers a detailed examination of the machine learning algorithms employed, the probabilistic reasoning framework, and the simulation results, providing valuable insights into the future trajectory of renewable energy growth and highlighting the transformative role of machine learning in energy forecasting.