Development of design system of coastal and river structures using neural network

利用神经网络开发海岸和河流结构设计系统

• 批准号: 06555148
• 负责人: MASE Hajime
• 金额: $ 3.58万
• 依托单位: KYOTO UNIVERSITY
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (B)
• 财政年份: 1994
• 资助国家: 日本
• 起止时间: 1994 至 1996
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-06555148/
• 关键词: neural network; rubble mound breakwater; armor layr; ニュートラルネットワーク; 消波ブロック; 護岸; 混成防波堤; 河川護岸; 天然河岸; 衝撃砕波力

The stability of rocky slope under wave attack have been investigated by mainly hydraulic model tests, and empirical formulae have been proposed based on the test results. Van der Meer (1988) proposed a formula for the rocky slope stability based on a large number of irregular wave experiments. However, the formula is insufficient to represent the damage level. Kaku et al. (1991) and Smith et al. (1992) newly proposed an empirical formula to improve the degree of agreement between the measured and predicted damage levels. The formula, however, is not universal ; which requires a change of the coefficients in the formula for different data source.It is desired to make a universal formula or a methodology to arrange the experimental data having ambiguous mutual relationships between physical factors (Mase et al., 1995). A neural network modeled on the structure of brain is effective to deal with information having not clear relations between causes and effects. This study shows the efficiency of neural network method to assess the stability of rubble mound breakwaters and armor layrs. In addition to show the efficiency, sensitivity of some parameters governing the stability (taking up the damage level as a parameter representing stability) is evaluated by utilizing the neural network where the number of acting waves, the stability number, and the surf similarity parameter are always employed to represent at least the characteristics of breakwaters and waves. For the case of rubble mound breakwaters, the permeability coefficient and the dimensionless water depth were found to be significant. However, for the case of armor layr, the difference of sensitivity of input parameters was little.

主要是液压模型测试，已经研究了岩石斜率在波攻击下的稳定性，并根据测试结果提出了经验公式。 Van der Meer（1988）提出了一个基于大量不规则波实验的岩石斜率稳定性的公式。但是，该公式不足以表示损伤水平。 Kaku等。 （1991）和Smith等。 （1992年）新提出了一种经验公式，以提高测得的损伤水平和预测损伤水平之间的一致性程度。但是，公式并不普遍。这需要更改公式中不同数据源的系数。希望制作通用公式或一种方法来安排物理因素之间具有模棱两可的相互关系的实验数据（Mase等，1995）。以大脑结构为基础的神经网络可有效处理因原因和影响之间没有明确关系的信息。这项研究表明了神经网络方法评估瓦砾丘的稳定性和装甲路线的稳定性的效率。除了显示效率外，通过利用神经网络的稳定性（将损害水平作为代表稳定性的参数占用的参数）的灵敏度通过利用表演波数，稳定性数量和SURF相似性参数为始终被用来至少代表防波堤和波浪的特征。对于瓦砾丘的防波堤而言，发现渗透系数和无量纲的水深很重要。但是，对于Armor Layr而言，输入参数的灵敏度差异很小。

项目成果

後藤仁志: "ニューラルネットワークを用いた護岸被災形態の類型化" 水工学論文集. 第38巻. 833-836 (1994)

Hitoshi Goto：“使用神经网络对海堤的损坏模式进行分类”，《水利工程学报》，第 38 卷，833-836（1994 年）。

Hajime Mase: "Evalnation of Artificial Armour Layer Stability by Newral Network Methoe" Proc. 26th Congress of IAHR. 341-346 (1995)

Hajime Mase：“通过 Newral Network Methoe 评估人工装甲层稳定性”Proc。

Mase, H.and Takayama, T.: "Evaluation of sensitivity of parameters affecting stability assessment of rubble mound breakwater" Proc.Japanese Conf.Coastal Eng.Vol.43, JSCE. 801-805 (1996)

Mase, H. 和 Takayama, T.：“影响碎石堆防波堤稳定性评估的参数敏感性评估”Proc.Japan Conf.Coastal Eng.Vol.43，JSCE。

Hajime Mase: "Neural Network for Stability Analysis of Rubble-Mound BreaKwaters" Jour. Waterway, Port, Coastal, and Ooan Engineering. Vol. 21, No. 6. 294-299 (1995)

Hajime Mase：“用于瓦砾堆防波堤稳定性分析的神经网络”杂志。

間瀬肇・高山知司: "ニュートラルネットワークによる捨石防波堤の安定性に及ぼす支配因の影響度評価" 海岸工学論文集. 43. 801-805 (1996)

Hajime Mase 和 Satoshi Takayama：“使用中性网络评估控制因素对碎石防波堤稳定性的影响”海岸工程杂志 43. 801-805 (1996)。