Comfort-controlled local personalized climatization using multimodal sensor data fusion and thermal comfort modeling

• Applicant: Professor Dr.-Ing. Christoph van Treeck
• Subject Area: Construction Material Sciences, Chemistry, Building Physics ; Structural Engineering, Building Informatics and Construction Operation
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 556342419

Personalized climate control systems (PCS) create a microclimate around the user and show significant energy efficiency potentials. The accepted target room temperature range can thereby be extended to 18 °C to 30 °C. However, a dynamic control system is required, which takes into account the building's purpose, thermal comfort and energy saving goals. The project aims to research a comfort-based control concept for personalized thermal conditioning in workplaces, which puts the user at the center of indoor climate control. For this purpose, a control strategy for local actuators based on sensor data and an individualized thermophysiology and integrated comfort models will be investigated and implemented. Building on conventional (PID) control strategies, the potential of advanced approaches using machine learning and model predictive control (MPC) is investigated. Input data includes, among others, a 3-camera system for recording and assigning local skin/surface temperatures, as well as an existing numerical thermophysiology engineering model (MORPHEUS), which determines the thermoregulation response of individuals in real time by adapting the model parameters to those using bioimpedance analysis (BIA) for body composition analysis. The aim is to couple individualized numerical approaches with locally measured values and a comfort model based on physiological input signals, e.g. skin temperature, which makes it possible to use the thermal comfort of individuals as a climate sum control variable in real time and taking dynamic effects into account. Based on the person-specific comfort statements of the virtual models, local climate actuators such as an air-conditioned office chair (heating and ventilation function), personalized ventilation, radiation panels, etc. shall be individually controlled. A bidirectional server-based communication approach via TCP/IP using MQTT is adapted for this purpose. Subject studies in a virtual office space in the applicant's new indoor climate laboratory depict four scenarios for moderate thermal conditions over four seasons and evaluate the approach and examine the different typological control approaches.

• DFG Programme: Research Grants
• Co-Investigators: Professor Dr.-Ing. Jérôme Frisch; Dr. Marc Syndicus