Risk Assessment For Ship Collision In Estuarine Waters Using AIS Data

Estuarine waters,as vital corridors,play a significant role in connecting inland and open sea shipping transport.With the rapid development of China's economy and shipping industry,the trends of ships' maximization and speed's incrementation become more tangible than ever,resulting in increasing traffic density and complicating the navigation environment in such areas,which pose increasing risk to the safety of the public's life,property and environment.To improve navigation safety level and economy development,it is of great importance to conduct researches on risk of ship collision accident.This study proposed a risk assessment method combining theoretical calculation and simulation analysis using ship AIS data and historical ship collision records.Western waters of Shenzhen port is chose as case study to validate the method.For the theoretical calculation,firstly a ship collision candidate detection model was established based on Fuzzy Quaternion Ship Domain(FQSD)theory.A causal probability model for ship collision was then proposed to perform Bayesian inference under influence of multiple influencing factors.For simulation analysis,a maritime simulation model based on artificial potential field was established to simulate vessel traffic and collision candidate.As case study,western waters of Shenzhen port were chose to apply the model established,the collision risk and its spatial-temporal distribution characteristics and etc.are analyzed to test the reliability and efficacy of the theoretical and simulation model.The main contents of the research are as follows:(1)A comprehensive literature review concerning current trends and methodologies applied in the fields of risk assessment of ship collision accident was conducted.Based on this,the main contents and knowledge of the methods this paper applied were also reviewed.(2)Three models for risk assessment of ship collision was established using AIS data and ship collision accident data respectively.Firstly,a collision candidate detection model based on fuzzy quaternion ship domain and ship movement are designed to obtain the number of collision candidate and its characteristics.Secondly,based on historical ship collision accident data,the influencing factors were analyzed in the terms of “human-ship-environment-management”.A Bayesian network model of ship collision were established by translating a fault tree based on causal chain of accident statistics to calculate causational probability of collision in the research area.A maritime traffic model on the basis of artificial potential field was finally established.Ship AIS data were introduced to obtain route characteristics using spectral clustering incorporated with Dauglas-Peucker compression as well as to obtain traffic parameters as input of the simulation.(3)Western waterways of Shenzhen port was chosen as case study.Number of collision candidate was obtained by applying collision candidate detection model using ship AIS data,as well as its spatial-temporal distribution and composition characteristics.The causal probability of ship collision and sensitive factors are obtained by implementing the Bayesian network model.A maritime simulation was also performed to compare the results with the collision candidate calculated using AIS data.Results of the comparison between simulation and historical data indicates that the method this study proposed can identify time span and water areas with high risk of ship collision as well as the sensitive causal factors of collision accident.The theoretical and simulation analysis method of ship collision risk can depict collision risk in estuarine waters using spatial-temporal distribution of collisoin candiate,ship type,and characteristics as well as the extent of sensitive influencing factors to ship collision.Risk of collision can also be quantified.Such results can be applied in the maritime safety management and design of port and waterways,and provide theoretical and technical reference to relevant policy management and engineering design.