Methods, data sources and applications of the Artificial Intelligence in the Energy Poverty context: A review

Energy Poverty (EP) is a widespread problem in Europe. EP detection is hampered by a lack of data and global metrics. Recently, innovative approaches using Artificial Intelligent (AI) techniques have been increasingly applied for the EP alleviation. In this work, studies focused on the application of AI on EP were studied. It was identified that there is not a high number of works that apply AI to fight against EP (considering this problem as a multidimensional phenomenon). Artificial Neural Networks-based algorithms and Decision Trees were the most used algorithms in the reviewed literature focused on EP alleviation. However, several AI applications focused on partial aspects of the EP or on areas intimately related to EP (low-income, high-energy price and low-energy efficiency of buildings) that allow the characterization of the problem in an efficient way have been published in recent years; the last 7 years published literature have been reviewed in this work. It was found that Neural Networks algorithms were the most used models for low-income, energy price and poor energy efficiency characterizations. Support Vector Machines-based algorithms were the most popular AI method applied on energy consumption related problems. Deep learning was the most popular technique for detecting energy billing irregularities and unpaid energy bills.

openalex: W4281665463