In cases of one or multiple faults in electric distribution systems, as well as in preventive maintenance, it is inevitable lack of power supply to some consumers. However, in both cases, after the isolation of the system sections in fault or under maintenance, it is possible to reconnect the healthful consumers without supply being located downstream of those sectors. For this it is necessary to obtain an appropriate and efficient service restoration plan, which promotes the restoration of supply to the consumers in the shortest possible time and without overloading any network equipment.Many methodologies have been proposed to determine service restoration plans in order to help the distribution systems operation. However, the application of the most of these methodologies is limited to small networks when compared to real networks that have thousands of buses and switches. This is because, when these methodologies are applied to large networks (with dozens of substations, feeders and hundreds of thousands of switches and buses), they are either unable to provide a real-time solution or reach the maximum capacity of computer processing. In order to overcome these limitations, other methodologies simplify the network representation, failing to consider all the network elements. In these cases, the solution provided for the simplified network may not have the same effect on the network in operation, which compromises the reliability of these approaches.This project proposes the development of a methodology to determine service restoration plans in large-scale distribution systems, without requiring any simplification in the computer representation of the network. The methodology should be able to provide service restoration plans in contingency situations or preventive maintenance, as well as be suitable to networks without automation or partial or fully automated. In order to do that, techniques like Evolutionary Algorithms, Multi-objective Optimization and Node-Depth Encoding will be used. It is important to highlight that, with this project, the student intends to continue his research that started in his master course, which has been supported by FAPESP (process number 2011/04714-2). This project will be developed in the Laboratory of Computational Analysis in Electric Power Systems, of the Engineering School of São Carlos, University of São Paulo.
News published in Agência FAPESP Newsletter about the scholarship: