System for identification and mitigation of problems in technical database of distribution companies using advanced machine learning techniques

• Grant number: 23/06655-0
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: December 01, 2023
• End date: November 30, 2024
• Field of knowledge: Engineering - Electrical Engineering - Power Systems
• Principal Investigator: Felipe Ribeiro Miranda
• Grantee: Felipe Ribeiro Miranda
• Company: Sinapsis Inovação em Energia S/S Ltda
• CNAE: Distribuição de energia elétrica ; Serviços de engenharia ; Pesquisa e desenvolvimento experimental em ciências físicas e naturais
• City: São Paulo
• Associated scholarship(s): 24/02926-2 - Enrichment of the database through the processing of advanced calculation applications for the use of a machine learning model applied to the solution for Identification and Mitigation of Problems in Technical Databases of Distribution Concessionaires, BP.TT; 24/06139-5 - Development and evaluation of machine learning models applied to the solution for Identification and Mitigation of Problems in Technical Database of Distribution Concessionaires, BP.TT; 23/14752-6 - Database enrichment through the processing of advanced calculation applications for the use of machine learning models applied to the solution for Identification and Mitigation of Problems in Technical Databases of Distribution Utilities, BP.TT; 23/15729-8 - Functional specification, testing and quality control of solution for Identification and Mitigation of Problems in Technical Database of Distribution Concessionaires Using Advanced Machine Learning Techniques, BP.TT

Abstract

Several sectors of the economy have been undergoing an intense process of digital transformation, supported by a series of innovative technologies developed in recent decades. All these processes and technologies have one thing in common: the dependence on consistent data, upon which disruptive applications can be developed. The distribution of electricity represents an essential first necessity service, demanding constant efforts to increase efficiency that can be transformed into energy supply with reduced costs, increasingly enabling access to the entire population. The management of the distribution system depends on a series of electrical calculations and other algorithms responsible for simulating the behavior of the system, subsidizing a series of investment and action decisions to ensure the supply of energy with the expected quality. From these functionalities, technical and non-technical losses of the system, current and future overloads, supply voltage levels, and other fundamental indicators for service management are derived. The regulatory agency itself expends efforts to calculate some of these indicators for regulatory remuneration purposes (e.g., loss indices). These calculations are data-intensive, requiring electrical, topological, and market information. Given the size and complexity of distribution systems and the still recent use of advanced and data-intensive technologies, the registration bases of distribution concessionaire companies still have relative precariousness, introducing relevant uncertainties in the calculation processes, thus reducing the efficiency of distribution service improvement efforts. There are currently products and efforts being employed to mitigate registration problems that improve calculations, but they are restricted to the definition of pre-established rules that take into account topological aspects (discontinuity of nearby network sections, etc.). This project aims to build an innovative solution that, based on a set of multi-nature data (technical registration, operational measurements, market, commercial processes, among others), employs advanced Artificial Intelligence solutions (Machine Learning, Deep Learning) with the aim of identifying and correcting non-easily observable registration problems or correctable with a pre-defined rule based on topology, but that significantly impact the simulation processes of the system's behavior for the various purposes related to its management. (AU)