Optimisation of Building Energy Management in the Tropics Using Artificial Intelligence Algorithms: Scientific Review

Abstract

Improving the energy efficiency of buildings is a major challenge for sustainable development, particularly in countries such as Benin where energy resources are limited while demand continues to grow. The objective of this article is to analyse and synthesize existing approaches for predicting building energy consumption, and to propose a theoretical framework for developing an artificial intelligence–based predictive model adapted to the local context. To achieve this objective, a systematic review of the literature was conducted, focusing on three categories of energy prediction methods: technical approaches, artificial intelligence (AI) approaches, and hybrid approaches. The methodological analysis highlights that while technical and hybrid models rely on thermodynamic equations, AI-based models use historical data to predict future energy consumption based on multiple environmental and operational parameters. The results of the review show that AI-based methods-particularly multiple linear regression, artificial neural networks, and ensemble techniques such as random forests-offer high predictive accuracy and greater adaptability, making them increasingly popular for building energy management. By examining environmental, socio-economic, and technological factors influencing energy use, the study proposes a structured theoretical framework integrating machine learning algorithms to improve prediction performance. The review also identifies research gaps and outlines a methodological pathway for developing an AI-driven predictive model tailored to Benin’s climatic and infrastructural context.