Towards an uncertainty evaluation model for marine track considering local sampling cloud and sample information.

Area of Science:

• Maritime Safety
• Navigational Engineering
• Artificial Intelligence in Maritime

Background:

• Assessing ship crew practical abilities during navigation, especially in complex port approaches, remains challenging.
• Traditional methods often rely on subjective expert judgment, lacking quantitative and objective evaluation metrics.
• The need for intelligent systems to evaluate navigational performance under uncertainty is growing.

Purpose of the Study:

• To develop a novel track evaluation model for assessing single ship inward-port navigation.
• To introduce a track belt division method to handle divergent track sampling data.
• To establish a comprehensive evaluation scheme integrating these methods for intelligent assessment.

Main Methods:

• Development of a track evaluation model using forward and backward normal cloud generators on a planar chart.
• Proposal of a track belt division method based on normal cloud drop contributions.
• Integration of the evaluation model and division method for a comprehensive sampling information-based scheme.

Main Results:

• The developed model effectively evaluated the practical abilities of crews during inward-port navigation.
• An example using M.V. DAQING 257 into Dalian Port demonstrated the model's practical effectiveness.
• Evaluation results showed consistency with subjective expert assessments, validating the model's reliability.

Conclusions:

• The proposed track evaluation model and track belt division method offer a robust approach for navigational performance assessment.
• This integrated scheme provides an intelligent evaluation method under sample information, addressing uncertainty.
• The findings contribute a new, data-driven approach to maritime safety and crew training evaluation.