摘要
分析了现在广泛采用的判定管涌破坏手段的不足之处。在分析广义回归神经网络的基本原理和算法基础上,建立了无黏性土管涌判别的广义回归神经网络模型。以前人试验结果作为对比,采用特征粒径和孔隙率作为判别指标,对土样的渗透破坏形式进行判别。计算结果表明,该模型的管涌渗流破坏形式判定结果与前人试验结果完全一致,该方法为无黏性土管涌渗流破坏形式的判别提供了新的研究思路。
We demonstrate the disadvantages of current judging standard for piping types. A generalized regression neural net- work model for evaluating the piping in non - cohesive soil is established, on the basis of the analysis of fundamental theory and algorithm of generalized regression neural network. Taking the previous studies as contrast examples, we judge the piping types in non -cohesive soil by adopting characteristic particle size and void ratio as evaluation indexes, and the results are in conformity with previous studies.
出处
《人民长江》
北大核心
2012年第1期42-44,90,共4页
Yangtze River
关键词
管涌
广义回归神经网络
无黏性土
流土
piping
generalized regression neural network
non - cohesive soil
flowing soil
