摘要
地铁车站等深基坑开挖施工中监测数据处理极其复杂 ,其经验多于理论。有鉴于此 ,本文提出了神经网络的处理模型 ,并用将遗传算法和BP最优化方法相结合所产生的一种高效率、高精度的算法来训练网络 ,并比较了各类训练网络方法的优缺点。由于测量数据存在误差 ,所以本文又着重分析了误差在网络中的前向传播与可控制性 ,首次提出了工程应用中网络的初值稳定性问题。最后对某地铁车站的开挖监测进行了实例分析。
The monitoring data analysis in deep foundation pit excavation is quite complex and it depends much on experience rather than on theory. In this paper, the model of genetic algorithm-BP artificial neural networks is proposed and the genetic algorithm is combined with error back propagation method to make an effective and accurate algorithm and optimize the neural network structure. The forward propagation and control of measured data error in the networks are analyzed, some solution is presented. At last, by analysis of case study some suggestions are given.
出处
《地下空间》
CSCD
2002年第4期290-293,共4页
Underground Space
基金
上海市重点学科 (岩土工程 )资助
关键词
遗传BP神经网络模型
深基坑
开挖
监测
测量
数据误差
数值稳定性
model of genetic algorithm-BP artificial neural networks
deep foundation pit excavation
monitoring
error of measured data
initial value stability
