Research On Intelligent Collision Avoidance Methods For Ships In Complex Encounter Situations

As an important means of transportation at sea,ships have always played an irreplaceable role.With the development of commerce and trade,water traffic is getting heavier,which poses a huge test for the safety of ships' water navigation.Similar to unmanned driving,unmanned driving and assisted driving of ships have become a research hotspot.Ship collision avoidance is a key step in the ship's "unmanned" operation.Research on ship collision avoidance and route planning can greatly reduce the probability of water traffic accidents,and promote the development of shipping regulations.This topic is mainly from the navigation environment and ship control,to study the problem of ship collision avoidance in complex encounter situations,reduce the occurrence of ship collision accidents,and improve the safety of water navigation.Firstly,the frame of ship collision avoidance system is established.This paper analyzes the characteristics of the complex meeting situation,then studies the process of ship collision avoidance and divides the meeting situation,and establishes the collision risk model based on DCPA and TCPA.Based on the analysis of the International Rules for Preventing Collisions at Sea,the rules for preventing collisions are quantitatively described as the control constraints,and the ship control model is established in combination with the ship's own operation performance.Secondly,the idea of clustering was introduced,the CFSFDP algorithm was used to classify the static obstacles,and the method of independent selection of clustering algorithm combination parameters was proposed.The accuracy of the improved clustering effect was improved by more than 30%,and the degree of autonomy was improved.For dynamic obstacles such as fleet,the dynamic grouping of sailing ships can be realized by grouping and clustering based on ship similarity and through simulation verification.After clustering,it can reduce the meeting situation,reduce the number of collision avoidance ships,and reduce the complexity of meeting situation.Then,the method of dynamic ship position prediction is studied.In the network training stage,k-means algorithm and Wolf algorithm are applied to optimize the parameters of the neural network.In this process,an adaptive step size improvement method of Wolf algorithm is proposed.Through simulation,the prediction accuracy of the neural network with improved Wolf pack algorithm is improved,and the fluctuation of prediction error is reduced and the prediction effect is more stable.Finally,the intelligent collision avoidance method is verified by experiments.According to the complex encounter situation,ship collision avoidance simulation is carried out for the same kind of multi-obstacle environment,multi-obstacle environment and multi-vessel encounter situation.The experiment compares the obstacle avoidance method based on the raster method and the obstacle avoidance method based on the intelligent algorithm represented by the genetic algorithm.The results show that the research method in this paper has advantages in path smoothness and algorithm efficiency,and plays an outstanding role in multi-ship coordination and collision avoidance,which has certain feasibility.