Application of data analysis techniques for characterization and estimation in electrical substations

With the continued growth of smart grids in electrical systems around the world, large amounts of data are continuously being generated and new opportunities are emerging to use this data in a wide variety of applications. In particular, the analysis of data from distribution systems (such as electrical substations) can lead to improvements in real-time monitoring and load forecasting. This paper presents a methodology for substation data analysis based on the application of a series of data analysis methods aimed at three main objectives: the characterization of demand by identifying different types of consumption, the statistical analysis of the distribution of consumption, and the identification of anomalous behavior. The methodology is tested on a data set of hourly measurements from substations located in various geographical regions of Colombia. The results of this methodology show that the analysis of substations data can effectively detect several common consumption patterns and also isolate anomalous ones, with approximately 4% of the substations being identified as outliers. Therefore, the proposed methodology could be a useful tool for decision-making processes of electricity distributors.