Poster Abstract: Integration of Physics-Based Building Model and Sensor Data to Develop an Adaptive Digital Twin

The development of digital twins to help evaluate a structure's performance is critical in ensuring that a structure performs safely as designed. Recently digital twins have been used to model structural systems, where they form a digital representation of the structure, and can be used to simulate its structural responses. However, traditional digital twins for structures cannot perform effective structural monitoring over the service life of the building, as they cannot yet adapt to changes in the environment and member properties (e.g., damage). To overcome these challenges, we introduce a new adaptive digital twin that integrates the responses from the physics-based structural model and real-time sensor data. By integrating the sensor data into our physics-based model we are adapting the model to reflect the current state of the building. The physics-based models are developed to simulate the structural responses of a building under various dynamic loading, which are then updated by the building's sensor data through transfer functions, forming our adaptive digital twin. We evaluated our digital twin using a real-world 9-story building where the error of the dominant frequency peaks and amplitudes were reduced by up to 55.97% and 33397.4%, respectively.