Assessment of Ship-Overtaking Situation Based on Swarm Intelligence Improved KDE

Area of Science:

• Maritime Safety
• Data Science
• Navigational Risk Assessment

Background:

• Ship overtaking presents significant navigational risks, necessitating accurate risk assessment models.
• Current methods may lack the precision required for dynamic maritime traffic analysis.
• Intelligent algorithms offer potential for improving the accuracy of risk prediction.

Purpose of the Study:

• To develop a data-driven risk assessment model for ship overtaking scenarios.
• To enhance the accuracy of predicting ship overtaking points and their probability distributions.
• To provide pilots with effective decision support for safe navigation.

Main Methods:

• Utilizing particle swarm optimization (PSO) to improve kernel density estimation (KDE).
• Employing a cost objective function based on minimizing mean square error for bandwidth optimization.
• Developing an improved adaptive variable-width kernel density estimator to prevent overly smooth probability density estimations.

Main Results:

• The proposed model accurately displays the probability distribution of ship overtaking points.
• Optimized bandwidth and density analysis provide a probability-based risk evaluation.
• The adaptive kernel density estimator demonstrates convergence and reduces estimation smoothing.

Conclusions:

• The developed model efficiently evaluates ship overtaking risk status.
• It offers valuable navigational auxiliary decision support for pilots.
• This approach contributes to safer maritime operations through intelligent risk assessment.