Development of Sensors and Software Integrated System for Identifying the Microclimate in a Smart Classroom

This article presents the design and commissioning of an educational laboratory testbed for building thermal microclimate identification, combining the functions of a real physical plant and a digital twin platform. A data acquisition system based on an ESP32 microcontroller and DS18B20 sensors is proposed, providing a continuous stream of synchronized indoor and outdoor temperature measurements. A browser-based HTML/JavaScript tool is developed for the primary processing of the acquired time-series data. The resulting temperature profiles are used for subsequent parametric identification of room thermal dynamics within a grey-box modelling framework, as well as for training a Takagi-Sugeno fuzzy model that serves as a sandbox environment for the synthesis of a predictive controller based on the Twin Delayed Deep Deterministic Policy Gradient (TD3) deep reinforcement learning algorithm. The proposed solution is fully reproducible at minimal hardware cost using exclusively open-source software.