The Application of BPN Neural Network to Allocate Victims of an Earthquake Disaster Evacuation System

• Main Authors: Chao-Jung Chen, 陳昭蓉
• Other Authors: Kuang-Hui Peng
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/68947136697656342987

碩士 === 國立臺北科技大學 === 建築與都市設計研究所 === 92 === The problem of evacuation when the earthquake occurred is considered as a Bi-level Programming. The first level is the shortest distance and the second level is the satisfaction. There are many researches about the selection of shelter in the earthquake. Based on the investigation; the study will construct a BPN neural network for location-allocation model of refuge through the feature of experience learning to train the system with ability of logical thinking. The model can be planed to evacuate and assign personnel swimmingly and equitably within space and time are limited. We take Taipei City for the subject of database, then combine previous research and ARC VIEW system to put Dali City into practice for examine. In an actual example, we will analyze research and confer with the application in the light of evacuation when the earthquake occurred. By running GIS system, we can get the space data then put it into neural network for training and testing. When the training system is done to put the outcome into ARCVIEW system displayed the shelter location structure and assignment.