基于神经网络的多向波作用下直立堤越浪量预测模型  

Prediction model for wave overtopping calculation on vertical revetment based on neural networks

作　　者：张晓鹏 季新然 汤修潮 李金宣[3] ZHANG Xiaopeng;JI Xinran;TANG Xiuchao;LI Jinxuan(Hainan Branch of CCCC Infrastructure Maintenance Group Co.,Ltd.,Haikou 570228,China;College of Civil Engineering and Architecture,Hainan University,Haikou 570228,China;State Key Laboratory of Coastal and Offshore Engineering,Dalian University of Technology,Dalian 116024,China)

机构地区：[1]中交基础设施养护集团有限公司海南分公司,海南海口570228 [2]海南大学,土木建筑工程学院,海南海口570228 [3]大连理工大学,海岸和近海工程国家重点试验室,辽宁大连116024

出　　处：《水运工程》2024年第2期16-21,共6页Port & Waterway Engineering

基　　金：辽宁省自然科学基金项目(2021-KF-16-05)。

摘　　要：海岸工程中常采用直立堤的结构形式,波浪作用下的越浪量影响因素众多,特别是考虑实际多向波的作用,目前并没有统一的计算方法。利用神经网络方法在计算回归问题上的优势,构建反向传播神经网络模型,采用物理试验数据构建训练集,建立多向波浪作用下直立堤越浪量的预测模型。通过与测试集、经验公式以及其他物理模型试验结果的对比,显示了本模型具有良好的预测精度,可满足工程设计需要。利用该模型进一步分析波浪入射角度和方向分布宽度对波浪越浪量的影响规律,弥补了试验数据不全造成的变化规律不明显的问题。The vertical revetment is a commonly used structural form in coastal engineering.There are numerous factors affecting wave overtopping under wave action,especially considering the actual multi-directional wave action,and currently,there is no unified calculation method.This paper utilizes the advantages of neural network methods in regression problems and constructs a BP neural network model.By using physical experimental data to build the training set,a predictive model for wave overtopping of sloping revetments under multidirectional wave action is established.Through comparisons with the test set,empirical formulas,and other experimental results,the model demonstrates excellent prediction accuracy and can satisfy engineering design needs.Furthermore,the model is utilized to analyze the influence of wave incident angle and directional spreading on wave overtopping,addressing the issue of indistinct variations caused by incomplete experimental data.

关 键 词：越浪量 神经网络 多向波 直立堤 

分 类 号：P753[交通运输工程—港口、海岸及近海工程] U65[天文地球—海洋科学]