【摘要】 以往的震害预测方法往往需要调查建筑物的详细力学特性参数,这对个别社区的范围是可行的,但对整个城市的震害预测,则需要耗费巨大的财力、物力,因此需要研究适用于城市群体建筑物的快速震害预测方法.提出基于范例推理的建筑物震害预测方法,该方法通过建筑物震害范例的表示、索引、检索和调整模型,并运用Access建立了建筑物震害范例库,实现了基于灰色关联分析确定属性权重和欧式距离、曼哈顿距离、模糊相似优先及神经网络等的4种检索方法来预测城市未知建筑物的震害结果.该方法能够在较少的投资和时间下,迅速给出城市的震害预测结果.最后将该方法应用于厦门市的建筑物震害预测,结果表明该系统具有良好的可操作性和结果可靠性.更多还原

【Abstract】 Existing earthquake damage prediction methods need detailed mechanical character data of structures.These data may be possible for in a block, but for a large city, they are not available because of enormous financial and material consumption. So, it is necessary to research rapid methods for predicting earthquake damage of group of buildings in a city. In this paper a new rapid earthquake damage prediction method was established which was based on the case reasoning. It put forward the representation, adjustment and retrieval model, in which Access data-base was used to represent and storage the cases of earthquake damage of buildings. Meanwhile, grey-relational theory was used to confirm earthquake damage factors weight of buildings,and the four theories (Euclidean distance, Manhattan distance, fuzzy preferred ratio and neural network) were used to find out the best similar case in the storage. The application of the method to the earthquake damage prediction of the buildings in Xiamen City shows that this method is easy for operation and reliable.更多还原