Study On Key Technologies For Ship Intelligent Decision Making For Collision Avoidance Based On Rules

The study on ship intelligent decision-making for collision avoidance has always been a focus and front subject and the key technology of intelligent ship in the field of maritime traffic engineering.There are many qualitative terms of the Convention on the International Regulations for Preventing Collision at Sea(COLREGs)and seafarers’ ordinary practice,which should be complied with by the intelligent decision-making support system to guarantee the evasive action in accordance with the navigation practice.The key issues of the system include the identification of encounter situation and action manner,the requirement of navigation rules,the effect of manoeuvrability on anti-collision performance,the reliability and practicality of the anti-collision algorithm and the corresponding outputs,as well as the establishment of reasonble anti-collision strategy.In view of this,the thesis focuses on the following aspects.In order to guarantee the rationality of the anti-collision algorithm and improve the intelligence,a quantified identification model of encounter situition is proposed.The identification results are used as constraints to limit the optimal direction of the path planning.Based on the quantitative interpretation of encounter types in COLREGs and previous study,the study analysed the characteristcs of encounter and motion,and the sensitivity analysis is applied to refine the judgement criteria for the purpose of integrity and uniqueness.The action manners including course alteration and changing speed for different encounter situation are identified based on the seafarers’ ordinary practice and the cost of avoidance.Finally,the identification process is realized by Boolean expression.In order to improve the reliability and efficiency of the anti-collison algorithm and integrate the navigation experience into the algorithm,an anti-collision avoidance algorithm including course alteration and speed reduction is propposed,which considers the effect of ship manoeuvrability.The former introduces the multiple population co-evolution searching scheme to overcome the premature convergence by traditional genetic algorithm,the constraint conditions,and fitness function are established by considering the requirements of the navigation rules,ship safety and the restoration to the original state,the navigation experience is also integrated into the algorithm by using nonlinear programming technology.The latter analyzes process of speed reduction for collision avoidance,the restoration time is determined by the DCPA judgment method,the engine speed is finally calculated by the numerical optimization method to confirm the restoration location and time.Finally,simulation results including head-on,large angle crossing,small angle crossing and overtaing situation demonstrate effectiveness of the anti-collision algorithm.In order to guarantee the coordination among evasive actions of different ships in multi-ship collision avoidance decision,the design of multi-ship anti-collision strategy under complex encounter situations is constructed based on the theory of synergetics and queuing.This method assumes the ships within a certain range as a transporation system,the anti-collsion decision is considered as the stochastic service that each ship is waiting to accept the service.As the system state is jointly determined by the state vectors of all ships,an optimization echelon is constructed to tease out the action priority based on space collision risk index and time collision risk index.The multi-layer coding genetic algorithm is adopted to generate evasive manoeuvres for the ships in the same optimization echelon simultaneously.Finally,simulation results of three-ship encounter and six-ship encounter demonstrate the effectiveness of this proposed algorithm.This stragety makes decision objects convert from individuals to a class of groups satisfying certain conditions,realizes the cooperation between ships,and reduces the blindness and antagonism of the decision-making scheme.Based on the COLREGs and the seafears’ordinary practice,the study on the technoloty of ship intelligent decision-making for collision avoidance in the open sea is conducted.A practical anti-collision algorithm is proposed from the perspective of the rationality of the decision objects establishment,the reliability of the decision-making model as well as the applicability of the output decision scheme.The effectiveness of the proposed model is verified by simulation under different traffic scenarios.Compared with the previous research methods,the anti-collsion paths obained by the proposed method are in line with the navigation practice and the coordination of the avoidance actions among ships is also ensured.This doctoral dissertation has certain theoretical significance and application value to the research work in the field of intelligent decision-making for collision avoidance.