Data-driven modeling of ship maneuvering motion using adaptive gridding-based weighted twin support vector regression

Nonparametric modeling is a commonly used data-driven method for modeling ship maneuvering motion, and its performance depends on reliable training data. The training data are usually collected from free-running model tests or full-scale trials, which inevitably contain noise and outliers. These noise and outliers can impair the prediction accuracy and generalization ability of nonparametric model. To improve the robustness of the nonparametric model, a novel adaptive gridding-based weighted twin support vector regression (AGWTSVR) is proposed. An adaptive gridding (AG) method is proposed for outlier screening and training data weighting. The AG method employs principal component analysis to map the training data into a two-dimensional grid. Subsequently, outliers in the training data are screened based on grid metrics, and initial weights are assigned to non-outliers. Additionally, a feedback weighted strategy based on training error is adopted to finely allocate weights to the training data. The dung beetle optimization algorithm is also employed to determine the optimal hyper-parameters of the proposed method. The AGWTSVR is employed for nonparametric modeling of simulation data (Mariner class vessel) and real measured free-running model tests data (KVLCC2 ship), respectively. The results demonstrate that the AGWTSVR is a robust data-driven modeling method for ship maneuvering motion.

• Record URL:
  • https://doi.org/10.1016/j.oceaneng.2024.118942
• Record URL:
  • http://www.sciencedirect.com/science/article/pii/S0029801824022807
• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/00298018
• Supplemental Notes:
  • © 2024 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies. Abstract reprinted with permission of Elsevier.
• Authors:
  • Jiang, Lichao
  • Shang, Xiaobing
  • Lu, Lingyun
  • Li, Bo
  • Zhang, Zhi
• Publication Date: 2024-11-1

Language

• English

Media Info

• Media Type: Web
• Features: References;
• Pagination: 118942
• Serial:
  • Ocean Engineering
  • Volume: 311
  • Issue Number: 0
  • Publisher: Pergamon
  • ISSN: 0029-8018
  • EISSN: 1873-5258
  • Serial URL: http://www.sciencedirect.com/science/journal/00298018

Subject/Index Terms

• TRT Terms: Data models; Maneuvering; Regression analysis; Ships; Vector analysis
• Subject Areas: Data and Information Technology; Marine Transportation; Vehicles and Equipment;

Filing Info

• Accession Number: 01927156
• Record Type: Publication
• Files: TRIS
• Created Date: Aug 13 2024 5:00PM