Machine learning heating–cooling energy forecasting tool under future climate scenarios

In line with European energy and climate objectives, forecasting techniques and machine learning methods are increasingly applied to support the prediction, analysis, and management of energy consumption in the building sector. The primary objective of this research is to highlight the value of deploying neural networks—specifically long short-term memory (LSTM) architectures—for managing and predicting shifting heating and cooling loads under future climate uncertainties. In contrast to studies that focus only on current weather conditions or measured historical data, this work investigates how different data-driven models respond to changing climatic conditions and evolving heating and cooling demands. The comparison of the three modeling approaches provides insight into their predictive accuracy, interpretability, and ability to capture nonlinear and temporal patterns in building energy consumption. In this way, the study supports the development of forecasting tools for climate-resilient energy planning and improved building energy management. The system under study is Building D of the Faculty of Engineering at the Aristotle University of Thessaloniki. The research was based on climatic and primary data, which are derived from simulation through the TRNSYS software. Then the methods used to derive results, through these data, are Regression and Neural Networks using the Python language. The results indicate that the LSTM model achieved the highest predictive accuracy, with R2 values ranging from 0.75 to 0.84 for heating energy consumption and reaching approximately 0.93 for cooling energy consumption, while the FNN model showed intermediate performance and the MLR model exhibited the lowest accuracy due to its limited ability to capture nonlinear and temporal patterns. Therefore, this study demonstrates the unique capabilities of sequential deep learning models as vital decision-support tools for long-term climate-resilient energy planning and the development of sustainable, resilient building energy systems.