Intelligent prediction and data analysis tools for energy consumption derived from digital currency mining

• Grant number: 20/10083-4
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Effective date (Start): October 01, 2020
• Effective date (End): September 30, 2021
• Field of knowledge: Engineering - Electrical Engineering - Power Systems
• Principal Investigator: Wallace Correa de Oliveira Casaca
• Grantee: Pedro Mark Bianchi
• Host Institution: Universidade Estadual Paulista (UNESP). Campus de Rosana. Rosana , SP, Brazil

Abstract

Machine Learning (ML) has been a very attractive tool to support economic plans and decision-making of companies, investors and public agents of electricity sector. In this context, an emerging research topic is the energy consumption derived from virtual currency mining, i.e., the so-called cryptocurrency mining. Such an issue has caused concern among governments and other players of energy industry, including countries with low energy demand historically but that have allowed the use of cryptocurrency mining technology on a large scale. Considering the perspective of expanding the cryptocurrency mining activity, building robust forecast models capable of providing a general consumption prediction, as well as a more accurate energy estimate is of paramount importance. Therefore, this research project addresses the problem of forecasting the energy consumption which arises from massive mining of cryptocurrency, whose payment validation requires specialized hardware equipment in order to properly work. Data Analysis techniques will be applied in an effort to analyze energy consumption data, as well as Machine Learning algorithms to predict the electricity demand resulted from cryptocurrency mining. Two well-established databases will be employed in our study: the former, from the University of Cambridge, and the latter, from the Digiconomist platform. Moreover, three ML models will be studied, prototyped and tuned: Support Vector Machines, Gradient Boosting and Random Forests. The computational apparatus to be developed in this research will be validated on the aforementioned databases, thus contributing to this particular field of study as well as providing new data-driven tools for supporting more effective energy plans in the energy sector.