社上水库原型观测分析中的三种优选模型对比研究  

Comparative study on preferred model of prototype observation and analysis of Sheshang Reservoir

作　　者：周清勇 胡国平[1,2] 洪文浩[1,2] 熊磊[1,2] 

机构地区：[1]江西省水利科学研究院,江西南昌330029 [2]江西省水工安全工程技术研究中心,江西南昌330029

出　　处：《江西水利科技》2017年第3期183-187,共5页Jiangxi Hydraulic Science & Technology

摘　　要：建立实用的大坝原型观测模型,能有效地拟合和预测大坝安全监测量.本文在土石坝监测量长系列观测资料分析的基础上,通过BP神经网络、逐步回归和多元回归进行拟合分析.结果表明,BP神经网络拟合效果优于逐步回归,逐步回归拟合效果优于多元回归.To establish the dam prototype observation model can effectively fit and predict the dam safety monitoring a- mount. Based on the dam long series of observations and monitoring analysis, BP neural network, regression and multiple re- gression analysis were fitted. The results showed that the fitting effect of BP neural network is better than stepwise regres- sion, stepwise regression is better than multiple regression.

关 键 词：大坝原型观测 BP神经网络 多元回归 逐步回归 模型对比 

分 类 号：TV698.1[水利工程—水利水电工程]