A Novel Method for Assessing Pirate Attack Risks and Spatial Distribution

Abstract

Pirate attacks pose one of the most severe challenges to the safety of maritime navigation. Effectively quantifying the risk of pirate attacks and understanding their spatial distribution through historical records is crucial for planning safe shipping routes. Given the diverse data types and multiple factors involved in assessing pirate attacks recorded in the Global Integrated Shipping Information System (GISIS) database, we propose a spatiotemporal influence factor analysis model based on the K-means clustering algorithm. Features are encoded using an Autoencoder, and the evaluation is conducted using the Entropy Weight Method- Technique for Order Preference by Similarity to Ideal Solution (EWM-TOPSIS). The model then simulates and predicts the geographical distribution of pirate risks. The results indicate that the model effectively captures the geographical distribution patterns of pirate attack incidents and successfully predicts the risk distribution across different sea areas. This approach aids in ship route planning and reduces the risk of pirate attacks.