圆弧滑动边坡反演设计的自适应神经模糊推理方法研究  

Study on ANFIS-based approach for inverse design of with circular failure surface sliding slopes

作　　者：丁德馨[1] 张志军[1] 

机构地区：[1]南华大学建筑工程与资源环境学院,湖南衡阳421001

出　　处：《岩土工程学报》2004年第2期202-206,共5页Chinese Journal of Geotechnical Engineering

基　　金：国家自然科学基金资助项目(50274043);湖南省自然科学基金重点资助项目(01JJY1004);湖南省教育厅重点资助项目(01A015)

摘　　要：圆弧滑动边坡反演设计的神经网络方法存在着收敛速度慢、拟合能力差、预测精度低、训练结果不具有唯一性等缺陷。针对这些缺陷,应用自适应神经模糊推理系统的原理,建立了圆弧滑动边坡反演设计的自适应神经模糊推理方法,并应用该方法对部分实例进行了反演设计。反演设计结果表明,该方法具有收敛速度快、拟合能力强、预测精度高、训练结果具有唯一性等优点,是一种优异的反演设计方法。ANN-based approach for inverse design of slopes with circular sliding surface has shortcomings such as slow speed of convergence, poor capability of fitting, low accuracy of prediction and indefiniteness of the training results. In order to overcome these shortcomings, Adaptive Neu-ro-Fuzzy Inference System is used to establish an ANFIS-based approach for inverse design of sliding slopes. Furthermore, this approach is used for the inverse design of several examples and the results show that it has the merits of high speed of convergence, good capability of fitting, high accuracy of prediction and definiteness of the training results, and is an excellent approach for inverse design of sliding slopes with circular fqilure surface.

关 键 词：圆弧滑动边坡 反演设计 自适应神经模糊推理系统 人工神经网络 

分 类 号：TU457[建筑科学—岩土工程]