Seminar 26 -- Integrating Data-driven and Physics-based Modeling Approaches for Advanced Research in the Built Environment PDF

2023 ASHRAE Annual Conference

This product is a .zip file that contains files that consist of PowerPoint slides synchronized with the audio recording of the speaker, PDF files of the slides, and audio only (.mp3 format) as noted.

Data-driven and physics-based models are commonly used for predicting building performance. Physics-based models require detailed knowledge of building characteristics and HVAC systems. Substantial efforts are needed to create and calibrate the models. A data-driven approach does not require detailed information on building characteristics or system specifications. However, data-driven models are sensitive to the amount of training data, data accuracy, and the operating conditions under which the data were collected. A hybrid modeling approach integrating data-driven and physics-based models can be flexible and efficient. We demonstrate the hybrid modeling approach through urban system modeling, model predictive controls and indoor greenery systems.