当前位置: X-MOL 学术Arab. J. Geosci. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Support vector machine for classification and regression of coastal sediment transport
Arabian Journal of Geosciences Pub Date : 2021-09-14 , DOI: 10.1007/s12517-021-08360-0
Mahdi Shafaghat 1 , Reza Dezvareh 1
Affiliation  

Predicting sediment transport rates has always been important in the field of coastal engineering due to its impact on coastline changes and port dredging. Today, in addition to experimental and quasi-experimental methods to study the phenomenon of sediment, the use of methods based on soft computing, machine learning, and artificial intelligence was considered. The reason is that the existing experimental methods presented completely different results and limited applications due to the significant influence of parameters and random data. In this study, the support vector machine is used as a new solution in modeling the sediment transport phenomenon (classification and regression). For this purpose, the coasts of Hormozgan province were selected in this study to classify and estimate the sediment transport rate. The coasts of this province are divided into western, central, and eastern parts, and a selected coastal segment is considered in each part. The purpose of classifying and estimating the sediment transport rate by the support vector machine is to predict this rate without the need for bed and beach profiles, as well as using the basic wave data. The classification results of sediment transport rate using the support vector machine method led to the selection of Gaussian kernel (RBF) with optimal coefficients of C=9 and σ=0.28, which indicated the appropriate classification of critical and non-critical states in each of the beaches by this machine. Further, the results of the support vector regression showed the appropriate estimation of the amount of sediment transport rate in all three coastal segments compared to the actual measured values.



中文翻译:

用于海岸带泥沙输移分类和回归的支持向量机

由于其对海岸线变化和港口疏浚的影响,预测泥沙输移率在海岸工程领域一直很重要。今天,除了研究沉积物现象的实验和准实验方法外,还考虑使用基于软计算、机器学习和人工智能的方法。原因是现有的实验方法由于参数和随机数据的显着影响而呈现出完全不同的结果和限制了应用。在这项研究中,支持向量机被用作模拟泥沙输运现象(分类和回归)的新解决方案。为此,本研究选择了霍尔木兹甘省的海岸对泥沙输移速率进行分类和估计。该省的海岸分为西部、中部和东部,每个部分都考虑了一个选定的海岸段。支持向量机对泥沙输移速率进行分类和估算的目的是预测该速率,而无需床和海滩剖面,以及使用基本波浪数据。使用支持向量机方法对泥沙输移速率的分类结果导致选择具有最佳系数的高斯核 (RBF)C =9 和σ =0.28,这表明该机器对每个海滩的临界状态和非临界状态进行了适当的分类。此外,支持向量回归的结果表明,与实际测量值相比,对所有三个沿海段的泥沙输移速率的估计是适当的。

更新日期:2021-09-15
down
wechat
bug