基于GRNN的砂土液化危害等级评价模型研究
Study on Hazard Degree Evaluation of Sand Liquefaction Based on the Generalized Regression Neural Network
作者：薛新华(1. 四川大学 水力学与山区河流开发保护国家重点实验室,四川 成都 610065；2. 四川大学 水利水电学院,四川 成都 610065)；陈群(1. 四川大学 水力学与山区河流开发保护国家重点实验室,四川 成都 610065；2. 四川大学 水利水电学院,四川 成都 610065)
Author：Xue Xinhua(1. State Key Lab. of Hydraulics and Moutain River Eng., Sichuan Univ., Chengdu 610065,China; 2. School of Water Resources and Hydropower, Sichuan Univ., Chengdu 610065,China)；Chen Qun(1. State Key Lab. of Hydraulics and Moutain River Eng., Sichuan Univ., Chengdu 610065,China; 2. School of Water Resources and Hydropower, Sichuan Univ., Chengdu 610065,China)
收稿日期：2008-11-12 年卷（期）页码：2010,42(1):42-47
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：广义回归神经网络；砂土液化；危害等级；评价
Key words：generalized regression neural network; sand liquefaction; hazard degree; evaluation
基金项目：其它
中文摘要
影响砂土地震液化的因素复杂且具有随机性和不确定性。神经网络方法不仅能考虑定量因素，而且能考虑定性因素的影响，因而神经网络方法适用于解决非确定性的砂土地震液化评价问题。在分析广义回归神经网络的基本原理和算法基础上，建立了砂土液化危害等级评价的广义回归神经网络模型。然后用收集到的工程实例样本对该模型进行训练和检验，并与BP神经网络判别结果进行对比。结果表明，广义回归神经网络性能良好、预测精度高，是砂土地震液化危害等级评价的一种有效方法。
英文摘要
The factors which control and affect sand liquefaction, are random and uncertain. Because artificial neural network can consider both quantitative and qualitative factors, it is suitable for solving uncertain problem of the evaluation of sand soil liquefaction. A generalized regression neural network model for evaluating sand liquefaction hazard degree was established based on the basic principles and algorithms, and was trained and checked with the collected sand liquefaction examples and compared with results obtained by the BP neural network. The results show that the generalized regression neural network model presents excellent network performance, high prediction accuracy, and is an effective way to evaluate the sand liquefaction hazard degree.
【关闭】
