Application of artificial neutral network and geographic information system to evaluate retrofit potential in public school buildings. (August 2019)

• Record Type: Journal Article
• Authors: Re Cecconi, F.; Moretti, N.; Tagliabue, L.C.
• Is Part Of: Renewable & sustainable energy reviews. Volume 110(2019)
• Page Start: 266
• Page End: 277
• Publication Date: 2019-08
• Subjects: Energy retrofit -- School buildings -- Open data -- Artificial neural networks (ANN) -- Geographical Information System (GIS) -- Data-driven process; Renewable energy sources -- Periodicals; Power resources -- Periodicals; Énergies renouvelables -- Périodiques; Ressources énergétiques -- Périodiques; 333.794
• DOI: 10.1016/j.rser.2019.04.073
• Languages: English
• ISSNs: 1364-0321

Abstract

Abstract: School buildings in Italy are outdated, in critical maintenance conditions and they often perform below acceptable service levels and quality standards. Nevertheless, data supporting renovation policies are missing or very expensive to be obtained. The paper presents a method for evaluating building's energy savings potential, using the Building Energy Certification (Certificazione Energetica degli Edifici - CENED) open database. The aim of the research concerns the development of a data-driven set of methods, based on the use of open data, machine learning (ML) and Geographic Information Systems (GIS) to support regional energy retrofit policies on school buildings. The main advantage concerns the possibility to predict the post-retrofit energy savings, avoiding the expensive on-site Condition Assessment (CA) phase. Data have been first clustered to identify the most common thermo-physical properties of the envelope, then three retrofit scenarios have been defined, to allow the retrofit of homogeneous types of buildings. The energy saving potentials have been evaluated through the implementation of eight Artificial Neural Networks. Ultimately, data have been geolocated and further processed to support the definition of the energy retrofit policies for the most critical regional areas. The Lombardy region has been chosen as case study to test the robustness of the proposed methods. The results of the case study proved that school buildings energy retrofit policies can be supported and defined using available open data, ML and GIS. The future developments of the research concern the further integration of GIS for retrofit cost assessment and scenario analysis. Highlights: Open-data, machine learning and spatial analyses support regional energy policies. Eight Neural Networks are used to compute energy savings in three retrofit scenarios. Data are geolocated and processed to guide the regional retrofit policy. The retrofit policy is defined avoiding expensive on-site Condition Assessment.