Font Size: a A A

Strategic Research For Ship Collision Avoidance Based On Automatic Identification System

Posted on:2019-04-29Degree:MasterType:Thesis
Country:ChinaCandidate:Z ZhaoFull Text:PDF
GTID:2392330590991886Subject:Naval Architecture and Marine Engineering
Abstract/Summary:PDF Full Text Request
Ship collision is the key factor of maritime traffic safety and maritime loss.With marine traffic become faster and denser,traditional ways of collision avoidance decision-making get higher pressure.In order to deal with this difficulty,ship automatic identification system became mandatory equipment on ocean vessels or large vessels,which could quickly gather information and make brief analysis.As the proportion of human factors in ship collision accident is too large,these information acquisition methods such as AIS have became the basis of research on automatic collision avoidance decision-making.At present,most ship automatic anti-collision methods make complete collision avoidance decision-making based on current situation and show the lack of consideration for the may occur in the process of dynamic collision avoidance.For more real-time collision avoidance decision-making method,reinforcement learning was select to search optimization sequential decision making with constantly updated information from automatic identification system.First,an absorbing Markov process is building by the beginning and end of the collision avoidance process.In order to meet the demands of continuous state space and continuous action space,an adaptive heuristic critic algorithm adopted radial basis function network was presented.In the algorithm,the differences between the ship navigation statuses are regarded as the input state of the Markov process,meanwhile the output decision is course alteration.Setting the return signals according to the risk of collision and course alteration,in which can reflect collision regulation.Finally,the algorithm will converge to ship collision avoidance paths and decision-making plans.The simulations in the MATLAB environment of ship collision avoidance paths were presented in different encounter situations,which conclude head to head situation,crossing situation,overtaking situation and multi ship encountering.Meanwhile,ship trajectories for other ships were also presented.Results show that give-way ships are complied with COLREGs and implement security anti-collision with the sequential decision obtained by adaptive heuristic critic algorithm.Compared with the existing distributed anti-collision decision,it shows better real-time operation.
Keywords/Search Tags:Adaptive Heuristic Critic, Ship Automatic Anti-collision, Risk of Collision, Reinforcement learning
PDF Full Text Request
Related items