Data analytics, modelling and optimization on HVAC systems

Abstract

This project explores the area of data analytics applied to Heating, Ventilation, and Air Conditioning (HVAC) systems, which leverage Time Series Forecasting, mainly the Darts library. The primary objective of this study is to develop and train accurate forecasting models for the key metrics produced by HVAC systems, which would eventually facilitate efficient energy consumption, and monitor the overall performance of these HVAC systems. To achieve this, two forecasting models, N-HiTS and N-BEATS are used. The performance of these forecasting models is continuously being compared across different metrics with the Mean Absolute Error (MAE), for us to find out which might be more suitable for which use cases. In conclusion, this project would contribute to the step forward in a data-driven approach for HVAC system management, bringing valuable insights to stakeholders to optimize the usage of their HVAC systems, reducing overall operational costs and improving efficiency.