Research

This research aims to develop a technique of building sensorization realized by analyzing the interrelation of structural responses among multiple buildings in a region for construction of urban safety network based on AI technology. From this research, the current concept of building safety evaluation focusing on an individual building will be expanded to a number of buildings in a city from a urban disaster management point of view.

As the building shows nonlinear behavior during the severe loads, so it is important to predict nonlinear structural responses to evaluate the status of buildings.

Data-driven approach using deep learning will be considered.

In this approach, the use of frequency-related data will be main point.

To do this, frequency domain data analysis methods such as continuous wavelet transform and seismic resonance area representing the relationship between buildings and ground motions in the frequency domain. 

For the precise safety evaluation of buildings, it is required to identify sensitive stiffness changes in buildings. So, to improve current damage detection method, deep learning and mathematical approaches will be employed to deal with scattered data from healthy and damaged states of buildings. In addition, through quantitative analysis for the data, some methods for damage severity will be developed. 

Finally, an integrated safety evaluation technique for buildings will be developed. In this research, the interrelation of structural responses for multiple buildings in a city is constructed using AI. And then, using this interrelation, those multiple buildings are going to predict and recover structural responses of neighbor buildings in case of sensor defects. Through this work, the current safety evaluation concept focusing on only one building can be expanded to the city and it will be urban safety management system.