Optimized Method for Locating the Source of Voltage Sags

Abstract

Short-Duration Voltage Variations (SDVVs) are the power quality disturbances (PQD) that mainly affect industrial systems, and are originated for various reasons, in particular short circuits over large areas, even those originating in remote points of the electrical system. The location problem aims to indicate the area or region or distance from the substation that is connected to the source causing the voltage sags, and is a fundamental task to ensure good power quality. One of the strategies used to determine the location of sources causing SDVVs and for an implementation of machine learning algorithms in modern distribution networks, called Smart Grids. Monitoring a Smart Grid plays a key role, however mostly it generates a large volume of data (Big Data) and as a result, multiple challenges arise due to the properties of this data such as volume, variety and velocity. This work presents an optimization through genetic algorithm to select meters which already exist in the Smart Grid, using a voltage sag location method in order to reduce the data obtained and analyzed throughout the localization process. Optimization was evaluated through a comparison with a non-optimized localization method, this comparison showed a difference between the hit rates of less than 1%.

Keywords

• voltage sag
• clustering algorithm
• genetic algorithm
• disturbance location
• smart grids
• big data