Early Warning Research On The Shipping Safety In Navigable Waters Of Coastal Ports

In recent years,with the implementation of the “going out” strategy,China's foreign trade has become increasingly frequent.As an important hub connecting the inland hinterland and overseas,the coastal port has continuously increased its port throughput and collection and distribution needs,and the shipping industry has shown a prosperous development trend.However,a series of safety problems such as an increase in traffic density in navigable waters and an increase in dangerous goods ships have become more apparent.According to statistics,due to the development of largescale ships,although China's marine traffic accidents have been slowly declining in the past decade,the direct economic losses caused by them are still very serious and remain high.In addition,with the introduction of a series of maritime safety policies issued by the State Council,China has increasingly strengthened its emphasis on maritime safety issues.How to accurately make advance warnings has become an urgent problem to be solved.In response to this phenomenon,this paper has carried out a comprehensive study on the safety warning of navigational waters in coastal ports.Firstly,based on the statistics of water traffic over the years,this paper analyzes the characteristics of accidents from the perspectives of accident development trend,type of accident,type of accident ship and time of accident,and introduces a security early warning system composed of early warning analysis subsystem and pre-control countermeasure subsystem;Secondly,through the analysis of the factors of water traffic safety,the total early warning index system of 4 categories and 27 categories is determined.In order to avoid its subjectivity,this paper takes Ningbo Port as an example,uses rough set(RS)to carry out the reduction of indicators and the determination of weights,and builds a four-category 16 small-category safety early warning indicator system with typical local characteristics;Then,based on the posterior probability idea,this paper constructs a four-class support vector machine(SVM)security early warning model based on the probability output of safe voyage,collision voyage,self-sinking voyage and aground voyage as the predicted state.The genetic algorithm(GA)is used to optimize the parameters of the SVM model.At the same time,this paper combines the probability of the model output with the loss weight by determined the analytic hierarchy process,and realizes the division of the security warning level from the urgency and the loss degree by means of MATLAB programming;Finally,in order to verify the accuracy of the model,this paper introduces the SVM model of grid search method(GS)optimization parameters and the GA-SVM model without rough set data processing for comparative analysis.It is concluded that the GA-SVM model has certain advantages in geometric accuracy and in a few classes,and the early warning performance of the model is reliable.The empirical analysis shows that the GA-SVM security early warning model can accurately predict the probability of accident type and provide early warning based on historical data,which can truly reflect the safety situation of the ship's navigation and achieve pre-existing monitoring and prevention of danger.This paper has a strong practical value.