Collision Hazard Identification of Unmanned Vessels in Inner River Based on Particle Swarm Parameter Optimization Support Vector Machine

Abstract

In this thesis, Collision hazard identification of unmanned vessels in inner river is studied. The electronic chart, radar and AIS system of ship assembly can provide the data that affect the normal operation of unmanned ships. The fuzzy comprehensive evaluation method is used to evaluate the collision of unmanned ships Levels of danger. The particle swarm optimization algorithm is used to support Vector machines classify the ship’s collision hazard classification. The data is based on data from 112 survey reports of collision between two ships issued by various provinces and local maritime authorities in China during 2004-2015 as a simulation sample. The collision risk of unmanned ships is divided into three levels, and Proposed a different level of unmanned ships coping methods. By comparing experimental results with ordinary support vector machine, particle swarm optimization, genetic algorithm and grid search parameters to optimize the performance of support vector machine, the support vector machine based on Particle Swarm Optimization can be used to predict the collision hazard identification of unmanned vessels, and it has a good performance.