边坡稳定可靠度替代模型分析

发布时间：2018-06-20 00:03

本文选题：边坡工程 + 替代模型　；参考：《中南大学学报(自然科学版)》2014年08期

【摘要】：首先对多项式响应面、BP神经网络、径向基函数及支持向量机这4种替代模型的基本理论、适用范围及构造程序进行具体分析;然后,针对BP神经网络、径向基函数及支持向量机这3种替代模型中训练样本不易获取的情况,引入拉丁超立方试验抽取样本数据,并通过相应替代模型构建边坡可靠度求解的稳定极限状态功能函数近似表达式;最后,通过工程实例对各模型的计算工作量及计算精度进行评估。研究结果表明:对于地层结构较简单、随机参数较少的边坡可采用多项式响应面替代模型,而对于地层结构复杂、功能函数具有较强非线性的边坡建议采用支持向量机替代模型,这可为边坡稳定可靠度替代模型的甄选与构建提供重要依据。
[Abstract]:Firstly, the basic theories, applicable scope and construction program of the four alternative models, namely polynomial response surface BP neural network, radial basis function and support vector machine, are analyzed in detail. In the three alternative models of radial basis function and support vector machine, the training samples are difficult to obtain, and the Latin hypercube test is introduced to extract the sample data. The approximate expression of the function of the stability limit state is constructed by the corresponding replacement model. Finally, the calculation workload and calculation precision of each model are evaluated by engineering examples. The results show that the polynomial response surface can be used to replace the model for the slope with simple stratigraphic structure and less random parameters, but the formation structure is complex. Support vector machine (SVM) is suggested to replace the slope model with strong nonlinear function, which can provide an important basis for the selection and construction of the slope stability reliability replacement model.
【作者单位】：湖南大学土木工程学院;湖南工程职业技术学院建筑工程系;
【基金】：国家自然科学基金资助项目(51278187) 湖南省教育厅项目(12C1025)
【分类号】：TU43

【相似文献】