Automatic identification of self-generation points in time series of electricity consumption : Granular Anomaly Detection

Abstract

The decrease in the prices of available technology for self-generation from solar energy, and the high environmental cost of traditional electricity generation systems, have led people in the context of climate change to generate their own energy to meet their consumption needs. For the electricity system at a strategic level, this has brought with it a series of challenges in terms of planning and projection of demand and its decreasing evolution over time, which suggests a technological challenge, especially when large cities or remote communities coexist. This article presents a methodology based on anomaly detection techniques for the characterisation of atypical changes in the behaviour of a time series of energy consumption, in order to identify the installation of self-generation devices by solar panels in a study area. The methodology analysed is based on mainly on two development trends : the first makes use of the anomaly detection algorithms available in the Prophet - Facebook library, while the second uses a series of exhaustive search algorithms to determine atypical changes in the data. The results obtained show the changes in the behaviour of the time series as a result of the integration of these technologies in electricity generation, and where the time interval of analysis plays a determining role in this process.