Researchers at the University of Missouri have developed an AI model that predicts college campus energy use with remarkable 1% accuracy. Utilizing six years of hourly data from the Combined Heat and Power Plant, Professors Sanjeev Khanna and Saad Alsamraee found that energy consumption barely decreased during the COVID-19 pandemic, challenging previous expectations. Traditional ARIMA forecasting models fell short, while their advanced machine learning techniques—including Decision Trees, Random Forest, and XGBoost—significantly improved prediction accuracy, with XGBoost reducing errors by 46%. This innovation allows campuses to optimize energy budgeting and planning, addressing variables like outside temperature that prior methods overlooked. Beneficial for universities and other facilities like hospitals and sports complexes, the model promotes cost savings and pollution reduction. Khanna emphasized its potential to enhance energy efficiency and reduce carbon emissions. With widespread application, this technology can revolutionize energy management for various institutions while contributing to climate change mitigation.
Source link
