J. Huete Cubillo, A. Elías Fernández, J. M. Morales González, S. Pineda Morente
Forecasting the energy consumption of buildings is essential to manage energy retailers' consumption portfolios, optimising RES, or improving energy efficiency. Nowadays, smart meters has enabled the access to consumption values in real time. The goal of this work is to exploit this information to forecast the energy consumption of multiple building one day in advance.
From a technical point of view, there are two major challenges: 1) The analysis of multiple fine-grained long time series with seasonal behaviours, and 2) the heterogeneity of the energy consumption patterns that might variate between buildings/consumers.
We tackle these difficulties from a functional data analysis approach, combining methods of functional time series forecasting with clustering procedures. Concretely, we explore similarity measures to cluster buildings based on functional envelopes.
To illustrate the methodology, we work with the energy consumption data of 499 customers located in Spain (year 2019).
Palabras clave: Forecasting, energy consumption, optimization, functional analysisProgramado
Sesión Invitada Math-In. Industrial Applications IV
9 de junio de 2022 17:10