Carbon Emission Prediction of Thermal Power Plants Based on Machine Learning Techniques

Abstract

Since the magnificent goal of Peak Carbon Dioxide Emissions and Carbon Neutrality was put forward in 2020, carbon emission reduction has attracted unprecedented attention. The power industry must fulfill its carbon emission reduction obligations as soon as possible. Thermal power plants are the main source of carbon emissions in the power industry, so finding out the key influencing factors of thermal-power-plant carbon emission and making accurate predictions are important measures to promote the low-carbon development of the power industry. Although some precise models have been proposed, most power plants cannot obtain all the parameters required by the precise models in the actual production practice, which limits their application. Machine learning technology accepts numerical data as input and establishes the mapping relationship between variables automatically, which results in loose requirements on data. This paper summarizes several key influencing factors of carbon dioxide emissions of thermal power plants that are easy to observe and establishes a prediction model of carbon dioxide emissions of thermal power plants based on eXtreme Gradient Boosting. In addition, we compare our method with two machine learning methods proposed in previous research and obtain a satisfactory result.