位移反分析的ANN方法与ANFIS方法的比较分析  

Comparison of ANN-Based with ANFIS-Based Approaches for Back Analysis of Displacement

作　　者：张志军[1] 丁德馨[1] 毕忠伟[1] 饶龙[1] 

机构地区：[1]南华大学核资源与安全工程学院

出　　处：《采矿与安全工程学报》2008年第2期159-163,共5页Journal of Mining & Safety Engineering

基　　金：国家自然科学基金项目(50274043);湖南省自然科学基金重点项目(01JJY1004);湖南省教育厅资助项目(05C491);湖南省安全生产科技发展计划项目(07-19)

摘　　要：各种传统位移反分析方法的反分析模型非常复杂,求解亦非常困难.而基于智能化理论所建的位移反分析方法的适用性仍需进一步研究和分析,基于此,对所设定的一标准弹塑性问题,采用ANN和ANFIS分别建立了位移反分析的ANN方法和ANFIS方法,进而又采用该标准弹塑性问题竖向位移的理论计算值分别应用这两种方法对其物理力学参数进行了反分析,并将反分析结果进行了比较.结果表明,ANFIS更适合于用来建立位移反分析方法.The model of traditional approaches for back analysis of displacements is very complex and it is rather difficult to solve them. The applicability of the approaches based on intelligent theory needs to be further studied and analyzed. So, ANN and ANFIS were used to estab- lish a new method, respectively, for back analysis of displacements. Furthermore, the two approaches were used to inverse the mechanical parameters for a prescribed elasto-plastic problem and a comparison Was made on the inverse design results. The research results show that the ANFIS is more adaptable to establish the ANFIS-based approach for back analysis of displacements.

关 键 词：地下工程 人工神经网络 自适应神经模糊推理系统 位移反分析 

分 类 号：TU413.6[建筑科学—岩土工程] TU457[建筑科学—土工工程]