Collapse Mode Generation and Prediction for Point Cloud BIM

  • 林 育正

Student thesis: Doctoral Thesis

Abstract

Taiwan is located in the Ring of Fire In recent years serious natural disasters have occurred frequently especially earthquake The seismic disasters result in structural failures and collapses and the disaster relief personnel must search and rescue (SAR) for survivors in damaged structures in a short time This study sets up a system to support SAR action that combines building information modeling (BIM) and dynamic simulation for building a database of the collapse structural model and presents a set of procedures for predicting the internal information based on external information in collapsed structures This study is divided into three major topics namely “Building information modeling” “Similarity algorithms” and “The Mathematization of database” The first theme uses the 3D computer graphics software Blender to make the building information model and simulates with discrete element method (DEM) then collect the results of different seismic periods to database The second theme applies Euclidean distance and cosine similarity to measure similarity then compare each results of similarities to find the most suitable BIM model for analysis The last theme applies artificial neural network (ANN) to mathematize the database of the collapse structural model then determine the feasibility of mathematizing with ANN The results showed the system can predict the internal information based on external information in collapsed BIM model and how to self-test the database and the ANN is useful to mathematize the database This study obtains good performance with reliability and efficiency because of great resolution capability and simple algorithms and processes
Date of Award2019
Original languageEnglish
SupervisorTsung-Chin Hou (Supervisor)

Cite this

'