Machine Learning Based Energy Management System for Smart Buildings

The requirement for efficient and sustainable urban environments has made energy management in smart buildings a crucial topic of research and development in recent years. This study targets gaps in energy management research by dynamically choosing the best ventilation devices (AC, fans, windows) based on the changeable environmental factors like temperature, humidity, and wind speed using reinforcement learning thus elevating energy management in smart buildings. The ML model was trained via reinforcement learning (RL), with a fuzzy control scoring system designed to take the consequent user satisfaction and power cost of the device selected by the model based on the weather conditions (temperature, humidity, and wind speed), to grade the selection during training. Moreover, verification of the model's decision-making mechanism was executed by an integrated simulation platform through case studies showing their applicability in typical and extreme conditions. The proposed system proved the capability of understanding the weather state to select the optimal device to turn on for the lowest power cost and highest user satisfaction in the building. This supports the United Nations (UN) sustainable development goals (SDGs) by promoting sustainable, cost-efficient energy use and improved living standards.