Research On Intelligent Evaluation Methods Of The Inbound/Outbound Single Ship In Ship-handling Simulator

Generally,traditional paper-based or subjective assessment approaches are adopted in many domestic evaluation institutions and universities in the seafarers' assessment examination of National Maritime Safety Administration.The evaluation results of which mainly depend on the subjective judgement of evaluators,leading to disunity of evaluation standard and heavy working intensity of evaluators,especially for the evaluation items of bridge resource management by using ship-handling simulator.In order to evaluate seafarers' operating ability more reasonably,intelligent evaluation algorithms and models are established.By using the process of single ship inbound/outbound port as evaluation objects,setting the trajectory evaluation of single ship inbound/outbound port as main approach,which also establishes arithmetic and theoretical foundations for constructing intelligent evaluation system of ship-handling simulator.By analyzing the compositions of uncertain information based on the routing plan,discussing the uncertain problems of navigation evaluation when single ship inbound/outbound port in ship-handling simulator,describing the intelligent evaluation modeling when single ship inbound/outbound port,and the intelligent evaluation modeling process based on big data,the dissertation explicit the intelligent trajectory evaluation research is the key part of the intelligent evaluation when single ship inbound/outbound port.By using subjective information in the condition of no objective sample data existed but only subjective information,two index weight calculation models of trajectory evaluation based on AHP and improved CHA respectively are proposed.Arming at the case of partial objective sample data existed,index weight calculation models of trajectory evaluation based on PCA,EM and RSM respectively are proposed.And a trajectory intelligent evaluation model of improved CDHA based on rough set is established,on the basis of synthesizing rough set theory,CDHA method and information entropy,which takes account of fuzziness and randomness of subjective information and incompleteness of objective sample data generated from subject information.Making full use of excellent trajectory sample information is an important approach to optimize the reference standard in evaluation model.Using applicable machine learning algorithm based on sample information is the important way to improve reference standard.Trajectory cloud generation algorithm based on small samples is proposed due to the infeasibility of large data,on the basis of which a trajectory evaluation model based on local-sampled cloud is established.Three reference trajectory region are divided according to analysis principle of normal cloud drop contribution,and the intelligent evaluation model based on objective trajectory information when single ship inbound/outbound port is formed.The Test validation indicates that the model have good performance of processing objective sample information.In the examination evaluation of Maritime Safety Administration,the same evaluator in different time evaluate the same object may gives different evaluation results,different evaluators evaluate the same object may gives different evaluation scores,in the process of evaluation,it is an effective way to solve the above problem by using the characteristics and evaluate results of evaluated trajectory as the reference basis.A trajectory cloud similarity evaluation model based on two-dimensional normal cloud using comprehensive information is established,with reference to features and evaluation results of evaluated trajectory,and combined with objective information of trajectory generated during the evaluation.For specific,CDHA trajectory intelligent evaluation model using comprehensive information is proposed based on analyzing the evaluation process of single ship inbound/outbound port using comprehensive information.One-dimensional normal cloud similarity algorithm based on corrected expectation curve is established by one-dimensional normal cloud feature test after introduction of cloud similarity approaches.Cloud parameter track similarity evaluation model based on two-dimensional normal cloud model is proposed after construct recommended track generation algorithm.Experiment shows the model can solve relevance problem of track evaluated and to be evaluated using comprehensive information efficiently.The function of intelligent evaluation module of single ship inbound/outbound port in ship-handling simulator is preliminarily realized in VC++ and tested together in the ship-handling simulator on the basis of Matlab software.The research of this thesis is practically significant to improve fairness and scientificity of evaluation examination results,promote product development of ship-handling simulator,and provide theoretic basis and technical support for further research of intelligent evaluations system in ship-handling simulator.