Caracterización del consumo eléctrico del sector sanitario en Castilla y León

Laburpena

Sustainable and responsible energy consumption requires the implementation of measures aimed at a more efficient use of energy. As far as electricity is concerned, a thorough knowledge of energy expenditure makes it possible to determine the presence of unnecessary consumption, to optimize electricity demand by reducing consumption or to carry out proper load management. Moreover, taking into account the liberalization of the energy market, the knowledge acquired can be used to improve negotiations with suppliers and the contracting of tariff parameters, thus reducing costs. In this Doctoral Thesis an energy analysis is carried out for large infrastructures, such as the buildings of the Health Department of the Castilla y León Regional Government, characterized by a large number of energy supply points (corresponding to 257 buildings) with high consumption and dispersion, constituting the analysis of the data a complex process that requires more sophisticated techniques than the usual ones. During this work, it is proposed to develop high-level computer tools to acquire, store and process data related to electrical energy expenditure, as well as its subsequent visualization by the end user. For this purpose, a data model is proposed that is applicable to the analysis of different types of energy consumption. In addition, data is collected on climatic variables and construction variables that influence energy expenditure and that will be useful in the analysis and interpretation of the electricity consumption profile and its estimation. Based on the data model, analysis tools have been developed to create simple visualizations and models to extract the knowledge inherent in the energy data. Using machine learning techniques, groups with energy consumption profiles of similar characteristics are created. The analysis of the profiles or curves of the electrical profile has been performed using clustering techniques, dimensionality reduction, regression analysis and neural networks. In this way, it is possible to determine which buildings behave similarly and which deviate from normal behavior. The models obtained allow estimating the monthly energy consumption of each building in relation to climatic variables. Based on future estimates of climatic variables, the building's energy expenditure can be predicted. The use of these techniques includes the use of regression analyses that integrate information on the evolution of the electricity consumption profile.