Conceptual prediction of harbor sedimentation quantities using AI approaches to support integrated coastal structures management

Mohamed T Elnabwy, Emad Elbeltagi, Mahmoud M El Banna, Mohamed Y Elsheikh, Ibrahim Motawa, Jong Wan Hu, Mosbeh R Kaloop

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)
2 Downloads (Pure)

Abstract

Sedimentation is one of the most critical environmental issues facing harbors’ authorities that results in significant maintenance and dredging costs. Thus, it is essential to plan and manage the harbors in harmony with both the environmental and economic aspects to support Integrated Coastal Structures Management (ICSM). Harbors' layout and the permeability of protection structures like breakwaters affect the sediment transport within harbors’ basins. Using a multi-step relational research framework, this study aims to design a novel prediction model for estimating the sedimentation quantities in harbors through a comparative approach based on artificial intelligence (AI) algorithms. First, one hundred simulations for different harbor layouts and various breakwater characteristics were numerically performed using a coastal modeling system (CMS) for generating the dataset to train and validate the proposed AI-based models. Second, three AI approaches namely: Support Vector Regression (SVR), Gaussian Process Regression (GPR), and Artificial Neural Networks (ANN) were developed to predict sedimentation quantities. Third, a comparison between the developed models was conducted using quality assessment criteria to evaluate their performance and choose the best one. Fourth, a sensitivity analysis was performed to provide insights into the factors affecting sedimentation. Lastly, a decision support tool was developed to predict harbors' sedimentation quantities. Results showed that the ANN model outperforms other models with mean absolute percentage error (MAPE) equals 4%. Furthermore, sensitivity analysis demonstrated that the main breakwater inclination angle, porosity, and harbor basin width affect significantly sediment transport. This research makes a significant contribution to the management of coastal structures by developing an AI data-driven framework that is beneficial for harbors' authorities. Ultimately, the developed decision-support AI tool could be used to predict harbors' sedimentation quantities in an easy, cheap, accurate, and practical manner compared to physical modeling which is time-consuming and costly. © 2022
Original languageEnglish
JournalJournal of Ocean Engineering and Science
Early online date2 Jun 2022
DOIs
Publication statusPublished online - 2 Jun 2022

Keywords

  • Sedimentation
  • Coastal modeling system (CMS)
  • Artifical intelligence (AI)
  • Integrated coastal structures management (ICSM)
  • Sensitivity analysis

Fingerprint

Dive into the research topics of 'Conceptual prediction of harbor sedimentation quantities using AI approaches to support integrated coastal structures management'. Together they form a unique fingerprint.

Cite this