University of São Paulo: USP creates technology that reduces time without energy in case of electrical grid failures
Who hasn’t experienced any inconvenience when the electrical power is interrupted after a storm affects the networks that feed our home? Generally, what we do in a situation like this is request the repairs from the company responsible for the energy supply, but depending on the severity of the problem, the repair time can be long.
Technology prevents fraud and interruptions in electricity
To reduce the duration of this uncomfortable period without energy, which can generate several socioeconomic impacts and fines for concessionaires, researchers from the School of Engineering of São Carlos (EESC) at USP developed a system that uses artificial intelligence to optimize network maintenance services. affected, helping the operators of energy distributors in making faster decisions. Part of the technology, capable of reducing repair time by up to 20%, is already being adopted by Companhia Paranaense de Energia (Copel).
The work was published by the Institute of Electrical and Electronics Engineers (IEEE), during the 2022 IEEE International Systems Conference (SysCon). Through the development of some algorithms – computer codes that perform a certain task – scientists have proposed software that takes into account different information to determine the most efficient way to solve the problem. During the assessment, the computer program considers the weather conditions of the locations, the position of origin of the maintenance teams, the state and level of congestion of the roads and roads used for locomotion, identifying possible obstructions, in addition to the history of repairs and problems. occurred previously. The technology, which took about five years to develop, even uses a system from NASA, the US space agency, to help estimate weather conditions.
“There was no algorithm capable of gathering all this data, interpreting it and transforming it into useful information, chewed, translated and in real time for the operator, who may be new to the job. Often, the professional needs to make a quick decision, but with so many variables, he cannot interpret them in the necessary time and ends up following a standard strategy, which may not be the most effective to solve the problem. That’s why it’s important to have different sources of information to speed up the repair and even predict a possible failure so that, if it really happens, the energy returns as quickly as possible. Our system will be the operator’s right arm”, says Henrique de Oliveira Caetano, a doctoral student in the Graduate Program in Electrical Engineering at EESC and one of the authors of the study.
History in game
Failures in the electricity distribution system can occur for a variety of reasons, including operational factors, equipment failures, or be caused by external problems, such as cyber attacks and extreme weather events. Knowing the history of these occurrences can help expedite repairs made by technical teams.
“Our algorithms analyze information related to failures and accidents that have occurred previously in a given region in order to propose the best solution for the occurrence being attended and for others that may occur in the future. The system is able to assess the regions that are most susceptible to flooding, the equipment or locations that have the most failures, which are the most recurrent problems, among others. With this data, knowing in advance that there will be a storm, for example, we can even move teams to strategic positions before a failure occurs, further optimizing maintenance and reducing the value of fines for the concessionaire”, explains Matheus Fogliatto, PhD student. in Electrical Engineering at EESC and one of the authors of the research.
As for the routes traced by the new software to guide the maintenance teams to the places where the problems occurred, the algorithms were customized with the objective of leaving people without energy as little as possible. Therefore, the codes define the best route based on the repairs needed to solve the faults, different from conventional GPS applications.
“Basically, the map apps used today plot routes considering only two points, the origin and the final destination, indicating the fastest route for a simple locomotion. However, in the case of repairing an electrical network, the scenario is different because inspections often need to be carried out at more than one point on the network, in different locations. This is exactly what our algorithms consider when calculating routes, as it may be necessary to stop at a specific point for additional repairs”, explains Luiz Desuó Neto, PhD student in Electrical Engineering at EESC and also author of the article.
Tests
To validate the technology, the scientists carried out several tests with computer programs that simulate different scenarios of the reality of a power distribution system, the possible failures that may occur, atmospheric variables, history, in addition to the estimated time for repair of each problem, always seeking to reduce the losses both for those who run out of energy and for the concessionaire.
Considering the standard time required to fix common problems, it was possible to reduce the time without power by about 20% using the technology. For the professor of the Department of Electrical and Computer Engineering (SEL) at EESC, Carlos Maciel, who guides the doctoral students involved in the work, one of the highlights of the study is the ability of the software to unite and merge different sources of information, providing more agile and accurate decisions by operators.
“From the customer’s point of view, it’s less time without energy, while in the eyes of the concessionaire, there will be much smaller fines to pay. Electric power distribution systems will fail, this is inevitable, even though equipment technology has evolved in recent years. So, we have to develop strategies to improve and accelerate the recovery of networks”, concludes the professor, who coordinates the Signal Processing Laboratory (LPS) at SEL, where the study was developed. Check out the research video below.
The work, which also had the participation of former EESC student Rodrigo Fanucchi, was supported by the Fundação de Amparo à Pesquisa do Estado de São Paulo (Fapesp) and the National Council for Scientific and Technological Development (CNPq), through the INCT-SAC, based at EESC.