Research On Ship Selection Of Port State Control Based On Principal Component Analysis And Extreme Learning Machine

Maritime safety supervision and management technology has always been an important research direction in the field of marine traffic engineering.And ship safety inspection is an important means to improve the safety level and transportation efficiency of marine traffic,which is conducive to the safety of life and property on ships and to prevent polluted waters.According to the nature of the jurisdiction in which the inspection is carried out,the ship safety inspection can be divided into two categories:flag state control and port state control.As an important supplement to flag state control,port state control has continued to develop since its birth in 1982,and has had a tremendous impact on the entire shipping industry.The memorandum of regional port state control that China belongs to-the Tokyo Memorandum of Understanding(Tokyo MoU)opened a new inspection regime(NIR)in 2014,and its selection mechanism is divided into two priority levels(Priority ? and Priority ?).This paper first introduces the new inspection regime of the Tokyo MoU,analyzes the characteristics of the NIR of the Tokyo MoU,and finds that the NIR of the Tokyo MoU does not have regulations on how to conduct the selection and inspection for the ships of Priority ?.And because of the past inspection regime did not divide the inspection level,so there is a lack of relevant quantitative research on the ships of Priority ?,and the past targeting model is difficult to meet the current requirements.In response to this problem,this paper proposes a port state control targeting model based on principal component analysis and extreme learn ing machine.Based on the port state control database of the Tokyo MoU,this paper obtained some of the port state control data for 2017,and obtained a random 100 inspection data for the ships of Priority ? by calculation.Among them,the ship selection indicators are mainly built with reference to the NIR of the Tokyo MoU,and then based on the principles of comprehensiveness and accessibility,the data in the database is used to enrich the ship selection indicators.In this paper,the Pearson correlation analysis is carried out on the sample data,and then combined with Kaiser Meyer Olkin measurement and Bartlett test,the analysis shows that there is a certain correlation between the ship selection indicators.In order to reduce the impact of correlation between indicators and reduce the sample dimension,sample data is processed using principal component analysis.Then this paper introduces the theory of extreme learning machine,fits and predicts the samples,and predicts the detention and deficiencies of the arriving ships by learning historical data to determine whether to board the ship.Finally,the results are analyzed,and the prediction results are compared with the actual results to verify the feasibility of the model.In view of the fact that the Tokyo MoU new inspection regime is difficult to divide the inspection order of the ships of Priority ?,the model avoids the deviation of subjective factors and considers the correlation between,parameters,which can provide decision support for port state control for ship selection.