Research On Maneuvering Strategy Of Inland River Autonomous Ship Sailing Safety

With the rapid development of science and technology,artificial intelligence has penetrated more and more deeply into all aspects of human life.Autonomous technology,as an important branch of artificial intelligence development,has had a significant impact on socio-economic development and national defense construction.In recent years,with the rapid development of technologies such as automatic control,the Internet of Things,and big data,ship-related environmental sensing technologies,communication and navigation technologies have also been widely used,providing a broad technical feasibility for the development of autonomous ship.Artificial intelligence has become an important direction for the future development of science and technology.The field of ships is starting an intelligent revolution.The combination of artificial intelligence technology and research in the field of ships is becoming closer and closer.The successful realization of autonomous driving will fundamentally change the traditional "man-ship-channel" mode of action,forming a "ship-channel" direct mode of action,thereby greatly improving the efficiency and safety of the shipping system.Inland waterway transportation plays an extremely important role in our country.Due to the special features of inland rivers,such as densely passing ships and narrow inland waterways,inland waterway safety accidents often occur.In view of the driving safety of autonomous ships,whether the application of autonomous ships to inland rivers can effectively solve the problem of navigation safety of inland river ships,so the research on the safety of autonomous ships in inland river is imminent.The current research mainly focuses on the collision avoidance and path planning of diverless ships.The research on the sailing safety of autonomous ships is mostly a qualitative analysis of the impact of human factors on autonomous ships,and there are also some literatures that have carried out risk assessments on autonomous ships;however,little research had done on the research of safety strategies for autonomous ships in inland river.Therefore,it is of great significance to study the sailing safety of autonomous ships in inland river.The research can provide a certain theoretical basis for the research on navigation safety of autonomous ships in inland river.The main contents and conclusions are as follows:1)By analyzing the characteristics of autonomous ship,further analyze the characteristics of the sailing risk of autonomous ship,and on this basis,explain the composition of the sailing risk of autonomous ship.Firstly,summarize the risk factors affecting the sailing safety of traditional ships by literature induction.The risk impact factors of autonomous vessel shipping safety are analyzed based on the typical accident scenarios of collision,aground,stranding,fire and explosion,and self-sinking in common ship accidents.Investigate the inherent factors that affect the sailing safety of autonomous ship in inland river due to ship-shore integration,and establish a framework of factors affecting the sailing risks of autonomous ship based on the ship-environment-control system,which laying the foundation for sailing safety measurement of inland autonomous ship.2)The coordinate axis is constructed based on the influencing factors involved in ship attributes,environment attributes,manage and control attributes,and a classification model for autonomous sailing scenes is constructed from the perspective of the coordinate system.An improved fuzzy C-means algorithm based on distance evaluation is proposed to realize the clustering of autonomous ship shipping scenes in inland river.The wine data set in the international universal UCI database,which is used to test classification and clustering algorithms,is selected to verify the algorithm,and the rationality and applicability of the algorithm were verified.The clustering results show that the algorithm is practical and effective.It not only can adaptively give the optimal number of clusters,but also can verify the effectiveness of clustering,and can quickly achieve the purpose of optimal clustering.3)Taking three typical sailing conditions,such as changes in sailing environment,encounters with other ship,and berthing of ship,as examples,and describing the safety maneuvering strategies under these conditions.In view of the ship's collision avoidance problem when the ship encounters in the channel,analyzing the maneuvering strategy in the encounter situation.Firstly,divide the encounter situation,and then measure the collision risk degree of the ship through these indicators: DCPA,TCPA and safe encounter distance.Based on the rules of inland river collision avoidance and the situation that ship will encounter,systematically analyze the collision avoidance strategies that the autonomous ship should adopt under the situation of two ships encounter and multiple ships encounter situation.For the analysis of the maneuvering strategy under berthing conditions,the ship's berthing process and the berthing precautions are analyzed in turn,and then propose autonomous ship berthing and unberthing maneuvering strategies.4)Taking the traditional ship accident case as an example,when multiple strategies were adopted in the corresponding scenario,the safety index corresponding to each strategy is calculated,and the difference coefficient CRITIC weighting method-TOPSIS model is used to evaluate various manipulation strategies.This demonstrate the construction process of ship sailing safety maneuver strategy set.The results show that the evaluation model intuitively and accurately reflect the optimal and worst of the ship's navigation safety maneuvering strategies,and then build the sailing safety strategies set of autonomous ship in inland river.5)Based on the analysis of the existing deep reinforcement learning method models,a combination of deep learning and deterministic policy gradient algorithms is used to establish a framework for autonomous ship sailing stratege based on DDPG algorithm.Combined with the actual situation of inland river sailing safety accident,from the perspective of typical scenes(obstacle avoidance)encountered in the course of inland river autonomous navigation,representative sections(including inland straight waterways,curved waterways,bridge areas,three forks waterways)of inland waterway were selected.Using Python software to conduct simulation research,and realize the process of learning and updating of autonomous ship sailing strategy.The results show that after a period of self-learning,the autonomous ship can learn excellent maneuvering strategies,update the ship's navigational safe maneuvering strategies in time,and perform well in the test environment.