Research On Ship Automatic Collision Avoidance Decision Based On Multi-Objective Genetic Algorithm

In order to study the problem of automatic collision avoidance decision-making for ships in open waters,and to ensure the safe navigation of ships.In the study of ship collision risk index,this paper has improved the CRI calculation method based on space collision risk index and time collision risk index,and solved the problem of "excessive over-calculation" when calculating the ship collision risk index.In order to improve the computational efficiency of ship collision risk index,BP neural network,GA-BP neural network and adaptive fuzzy neural network were used to predict and analyze CRI.In order to ensure the rationality of the collision avoidance decision,this paper establishes the fitness function based on safety,economy,collision avoidance rules and opportunity of collision avoidance.The specific research results of this paper are as follows.(1)The ship collision risk model based on the risk of space collision and the risk of time collision.When neither of them is 0,the maximum value of the two is taken as the ship collision risk under the situation.The calculation result of the ship collision risk is "over-large",which is easy to cause misunderstanding in the collision avoidance decision.Based on this,this paper has made some improvements to avoid the phenomenon of "over-large" and make the calculation results more in line with the actual situation.Furthermore,the BP neural network is optimized by genetic algorithm and adaptive fuzzy algorithm to calculate the collision risk.In order to avoid the network that is calculated by DCPA and TCPA cannot cover all the factors affecting CRI,the randomly generated speed(V),heading(C),relative azimuth(Tr)and distance(D)are used as the input of the network to predict CRI,the results show that the adaptive fuzzy neural network prediction results are better.(2)Identify key collision avoidance ships.In order to deal with the problem of multi-ship encounters,this paper simplifies it and avoids collisions by avoiding the "focus on avoiding ships".The model established in this paper is that when there are multiple target ships around the ship,the ratio of the collision risk of each target ship to the sum of the total ship collision risks is taken as the dangerous degree of the target ship.The ship can be prioritized to avoid the target ship at a high level of danger.(3)Construct a multi-objective decision genetic algorithm.In this paper,fitness functions based on safety,economy,collision avoidance rules and opportunity of collision avoidance are established respectively:f1(x),f2(x),f3(x),f4(x).Because of the contradiction of each objective function,it is impossible to achieve the optimal solution of each objective function at the same time.In view of this,according to the characteristics of each objective function,the weight method and the constraint method are comprehensively considered to solve the problem.Therefore,the research of automatic collision avoidance decision-making for ships based on multi-objective genetic algorithm is constructed.(4)Simulation.The simulation experiment in this paper uses Matlab2018b as the experimental platform.The simulation experiments were carried out under the situation of head on situation,crossing situation,overtaking situation and multiple ships meeting situation.The simulation results show that the ship collision avoidance decision based on multi-objective genetic algorithm can realize the automatic collision avoidance between ships quickly and effectively.