Research On Collision Avoidance Assistant Decision Making Under Multi-ship Ecounter Situation

The gradual implementation of the great strategic concept of "Maritime Silk Road of the 21 st Century" has greatly promoted the rapid development of the world shipping industry,and also puts forward newer and higher requirements for ship theory and maritime safety navigation technology.At present,the theory of collision avoidance between two ships is relatively mature,but there is still not enough theory to support the research on collision avoidance decision making under complex situations such as multi-ship encounters.This paper studies the collision avoidance decision making problem under the multi-ship encounter situations combining with the International Regulations for Preventing Collisions at Sea.The collision avoidance mechanism of the ship is analyzed.According to the maritime rules,the multi-ship encounter situation is divided,and a new ship domain model is established.The model comprehensively considers the factors such as deck officers' capability,the size and maneuverability of ships,sailing area environment and other factors that affects the size and shape of the ship area,and sets the corresponding safe distance to approach for different encounter situations to ensure the reasonable determination of the time to avoid collision and the time to recovery route.Combined with the ship area and dynamic boundary theory,the fuzzy mathematics method is used to establish the quantitative calculation model of ship collision risk index.The grey relation analysis method based on entropy weight is proposed to qualitatively evaluate the ship collision risk index.Based on the principle of geometric collision avoidance,the quantitative calculation model selects five indicators,including distance to closest point of approach,time to closest point of approach,relative distance,relative azimuth and the ship speed ratio,as the ship collision risk evaluation indexes to calculate the value of ship collision risk.The qualitative evaluation model calculates the entropy value of each index factor,assigns a certain entropy weight to the grey relation coefficient in the grey relation analysis method,and obtains the gray relation value of each target ship by the product of the two,thus determining the ship's key avoidance target ship.The particle swarm optimization algorithm is applied to the collision avoidance decision problem in the multi-ship encounter situation.Combined with the annealing strategy in the simulated annealing algorithm,the improved particle swarm simulation annealing optimization algorithm is proposed.The objective fitness function of multi-ship collision avoidance is established after analyzing the multi-ship collision avoidance decision process in detail.The objective function comprehensively considers the factors including the lowest collision risk of each target ship,the smallest steering angle and the shortest recovery time.The probability feature of sudden change of the simulated annealing method is applied to the particle optimization process.A certain probability determines whether or not to accept the latest solution as the new current state according to the Metropolis criterion,thereby ensuring that the objective function has the best search results in the global area.At the same time,the shrinkage factor is introduced to replace the inertia weight coefficient in the velocity iterative formula of the particle swarm optimization algorithm,which eliminates the boundary limit of the algorithm.