基于人工神经网络的高层建筑结构选型知识发现  

ANN-based knowledge discovery for form-selection of high-rise building structures

作　　者：吕大刚[1] 张世海[2] 刘晓燕[3] 王力[1] 

机构地区：[1]哈尔滨工业大学土木工程学院,黑龙江哈尔滨150090 [2]南阳理工学院建筑工程系,河南南阳400047 [3]哈尔滨工业大学计算机科学与技术学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2006年第B07期351-355,共5页Journal of Harbin Engineering University

基　　金：国家自然科学基金资助项目（59895410,50378030）.

摘　　要：目前结构工程实例积累越来越丰富，而解决结构选型这一实际工程问题的知识却相对贫乏，因此将知识发现技术应用于工程结构的设计过程，可以有效地辅助设计决策过程，提高工程设计的智能化水平．分析了根据结构实例库进行知识发现的可行性，提出了改进的BP人工神经网络学习算法，开发了高层建筑结构实例库及其管理系统，将实例库系统提供的实例分别作为神经网络系统的训练样本和测试样本，对3层多输入多输出BP网络进行了训练和学习，工程实例应用结果表明：所提出的方法和所开发的系统可以发现实际工程高层建筑结构的选型知识，系统具有较好的自适应能力．At present, there are more and more design cases in structural engineering with the development of building industry, however, there is lack of knowledge of solving the form-selection problems for large-scale and complex structures. Therefore, the applications of knowledge discovery in database （KDD） in the design process of engineering structures can efficiently aid the design of decision-making procedures, and greatly improve the intelligent design levels. In this paper, the feasibility of knowledge discovery based on structural casebase is firstly analyzed, and then, an improved learning algorithm for BP neural network is put forward. A casebase of high-rise building structures and its management system are developed. Taking the cases from the casesbase system as training and testing samples respectively, a three-layer BP neural network is trained and learned. It is shown by a practical engineering case that some new form-selection knowledge can be discovered from the high-rise building structural casebase by the proposed method, and the developed KDD system has good self-adaptability.

关 键 词：改进的BP网络 知识发现 结构选型 实例库 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]