Study On Ship Collision Avoidance Decision-making Based On Cellular Particle Swarm Optimization

With the development of shipping industry, the maritime traffic accidents are also increasing, and the survey shows that the main reason of ship collision is the wrong collision avoidance decision-making of the crew. So it is very important to establish reasonable and effective collision avoidance scheme. In order to improve the safety and reliability of the ship navigation on the sea,the Cellular Propagation Particle Swarm Optimization(CPPSO) is applied on ship collision avoidance,and collision avoidance decision problem is considered as a problem of optimization search. The ultimate goal of search is to find an optimal steering angle,so as to ensure that the collision accidents could be effectively avoided.Particle Swarm Optimization(PSO) is a simple and easy to implement method,which has the disadvantages of low convergence speed and easy to fall into local optimum. This paper introduces the information transfer mechanism of the propagation of Cellular Automaton(CA)into the PSO algorithm. It is mainly about CPPSO and its application in the ship collision avoidance decision-making. The main work is listed as follows:1. In the article, we summarized the basic knowledge of ship domain, and made a specific description of the solving method of the motion parameters of ship, the importance of collision risk and the collision avoidance decision-making. The basic principle and the realization process of the basic PSO and the related knowledge of the CA were introduced,then we stated the Cellular Particle Swarm Optimization algorithm, which contained the Internal Cellular Swarm Optimization algorithm and the External Particle Swarm Optimization algorithm.2. To improve the optimization performance of PSO, an improved method whose name was CPPSO was proposed, which was based on the information transfer mechanism of the propagation of CA. Firstly,we analyzed the idea of the improved algorithm and the convergence of CPPSO from the aspects of the mathematics. Then in order to verify the performance of the algorithm, we did some numerical simulation experiment for the standard testing functions. The experimental results, not only did it show the effectiveness of the CPPSO, but also showed that the CPPSO had a higher search quality, especially its concentration and better stability.3. The paper considered the steering angle in the implementation of the ship collision avoidance action as a problem to be optimized,which solved the optimal steering angle by means of the CPPSO. The ship collision avoidance object function was established from the aspects of the security which represented the collision risk, the economy which contained the sailing time and distance and the route smoothing. Finally, with the CPPSO, the emulational experiments of the ship's three encounter situation were carried out, and we made a comparison with the Genetic algorithm. It showed the CPPSO was effective and feasible in the ship collision avoidance decision-making.