Energizados

Inter-American Development Bank

Energizados is a machine learning-based tool that detects non-technical energy losses, identifying abnormal consumption patterns to improve energy efficiency and reduce costs. Developed to address high energy losses in Latin America and the Caribbean, this tool allows entities to plan more efficient inspections and inspections, contributing to sustainability and the reduction of energy costs.

Energized improves efficiency in the detection and management of fraud and non-technical energy losses, allowing energy companies to optimize their resources and significantly reduce operating costs.

Uses machine learning to detect and reduce non-technical energy losses, such as electrical fraud. Analyzes abnormal consumption patterns to identify fraud. It combines three models for fraud prediction: a supervised model, a semi-supervised model and an analytical rules model. It uses monthly consumption data as the main input for the models. It includes a user interface for visualization and analysis of the detected data. It has shown effectiveness by increasing the capture of electrical fraud by 1.65 times in tests carried out.

Leverages open-source Python libraries, ensuring an architecture grounded in open standards. Utilizes machine learning algorithms such as LightGBM and Catboost to enhance prediction accuracy. Employs libraries like Matplotlib, Seaborn, and Pandas to facilitate data analysis and visualization. Embraces interoperability principles, enabling seamless integration with other systems and open data formats.