高层建筑结构实例库及其数据挖掘技术
论文作者： 涂庆
院 系 ： 土木工程学院
专 业：
工程力学
第一导师： 刘晓燕
英文题目： HIGH-RISE BUILDINGSTRUCTURE CASE BASE AND DATA MINING
学 位： 工学硕士
中文文摘： 当前，国内外高层建筑结构发展迅速，在结构型式、设计与建造方面，都积累了可供借鉴的丰富实例。本文收集整理了国内外一千余个高层建筑实例，建立了高层建筑结构实例库，研究了基于实例库的数据挖掘技术。主要内容有：1. 从关系数据库的理论出发，设计了高层建筑结构实例库的关系模式，并在Microsoft SQL Server 2000的开发环境下建造了实例库，录入了上千个实例数据，在Delphi 6.0的开发环境下建造了实例库的管理系统，管理系统的主要功能包括数据录入、修改、删除、查询和统计。2. 在深入分析实例库的数据特征的基础上，研究了值域规范化问题，提出了基于同义词典和值域词典的动态管理方法，并通过实例加以说明。3. 针对高层建筑结构实例库中数据的特点，对挖掘关联规则采用的Apriori算法作了改进，对关联规则的评价指标?支持度的计算方法进行讨论，提出了相对支持度的概念；研究了连续型数据的离散化问题，提出了基于属性之间制约关系的动态划分法，并通过实例加以说明；描述了定量型关联规则发现的全过程；提出了依据信息扩散原理的区间模糊划分法及基于信息量的统计方法，并给出模糊关联规则的发现过程；给出了三种关联规则后处理的方法：对关联规则在工程上的应用进行了探讨。4. 针对高层建筑实例库中数据的特点，修改了构造决策树的ID3算法；描述了基于高层建筑结构实例库，分别采用决策树和粗糙集的方法，产生分类规则的全过程；给出分类规则的后处理方法；比较了三种数据挖掘方法，给出它们的应用条件。
英文文摘: Now the progress of high-rise building structure at home and abroad is very rapid. Plenty of cases used for reference have been gotten together in aspects of form, design and construction structures. Over one thousand high-rise building cases have been collected and arranged at home and abroad to construct high-rise building structure case base, and Data mining based on case base has been researched into.The main contents in brief are as follows:1. Based on the theory of relational database, relational modules of high-rise building structure case base are designed in this paper. In the developing circumstances of Microsoft SQL Server 2000, case base is constructed which has thousands of case data. And in this paper, case base management system is constructed in the developing circumstances of Delphi 6.0, whose main functions include inputting, updating, deleting, querying and counting data.2. By analyzing data feature in the high-rise building structure case base in-depth, in this paper, the value field standardization question is made researches on, and dynamic management method founded on thesaurus and value field dictionary is come out with, moreover, cases are brought forward to explain the question.3. Aiming at data feature in the high-rise building structure case base, in this paper, the Apriori algorithms is improved on to discover association rules, and the compute method of support which is as a evaluation index of association rule is discussed, meanwhile, the concept of relative degree of support is put forward. By making researches on the question about dispersing continuous data, dynamics partition method on the basis of restricted relation between attributes is given and cases are adduced to explain it. The whole process of discovering ration association rules is described in this paper. Then, the fuzzy partition method according as information diffuseness theory and statistical method on the basis of information quantity is posed. The entire course of discovering fuzzy association rules is also described. Furthermore, the later stage disposals of three types of association rules are raised, and the application of association rules in the engineering project is approached.4. In connection with data feature in the high-rise building structure case base, the ID3 algorithms to construct a decision tree is advanced in this paper. The whole process of classification rules that are discovered by decision trees and rough sets is also described, based on the high-rise building structure case base. Then the later stage disposals of classification rules are brought forward. At last, three data mining methods and gives applications condition are compared in this paper.
中文关键词： 高层建筑结构、实例库、数据挖掘、关联规则、分类规则、决策树、粗糙集
英文关键词： 高层建筑结构、实例库、数据挖掘、关联规则、分类规则、决策树、粗糙集
