Research On Knowledge Assist-based Ship Survey Guidance Pattern And Its Key Technologies

Shipping is an important way of transportation.Ship is an important maritime vehicle for transportation in shipping industry,and the quality of ship is closely related to safety,efficiencyand environmental protection.As an important technical way to ensure ship's quality,ship survey plays a critical role in ship's life cycle.As the main participant of ship survey activity,the surveyor's personal ability has a great impact on the quality of ship in survey.In China,with a large scale increase of surveyor group,the technical skills of the individual vary considerably.For those who are new ones at work,they are lack of practical experience and learn of survey specifications,resulting in low quality of survey task.Hence,it is easy to see that survey knowledge assist is really necessary for surveyor in survey activity,as it makes great significance to improve quality and efficiency of the survey.At present,domestic and international scholars have carried out studies on ship survey,and all works focus on the fields of education and training,management method and survey task execution method.The study on surveyor-oriented and knowledge assist-based way has not been carried out yet.Most of the ship survey agencies are still using traditional survey rule retrieval system.Although there are fully-fledged theories and applications in fields of product design and product manufacture,they cannot be used to provide survey knowledge assist as the particularities of ship survey industry.In regard to above questions,the dissertation focuses on surveyor groups,in order to realize the survey process-based knowledge assist,the main concept and functional framework of ship survey guidance pattern(SSGP)are proposed.On the basis of that,three key issues in it,including knowledge modeling,knowledge acquisition method and task processing method,are discussed.Finally,the feasibility of the method is verified via a SSGP prototype.The main contents are as follows:1)Research on the design of SSGP.Aiming at current knowledge assist systems cannot meet the knowledge requirement of ship survey,and combining with the applications of guidance technology in other industries,the concept of knowledge assist-based SSGP is proposed.Meanwhile,two types of surveyor's knowledge requests in survey action are summarized,and the approaches to get the answers in SSGP are discussed.At the same time,the hierarchy and main functions of knowledge assistare briefly analyzed.2)Research on multi-domain knowledge modeling of ship survey.Aiming at the problem of ship survey knowledge description in traditional knowledge modeling way,technical information knowledge,organizational structure knowledge,business resources knowledge and business process knowledge in ship survey are analyzed.Combing with the knowledge characteristics,such as wide range domain,fulfill of hidden knowledge and inner close relations,based on semantics and semantic web and ontology theory,the method of ship survey knowledge modeling based on domain ontology is proposed.On the basis of this,knowledge in ship survey domain ontology(SSDO)is composed of four dimensions: survey object,survey rule,survey accessory and survey task.Hence,ship survey knowledge is described in detail and accurately by those four dimensions from different perspectives.Via the establishment of semantic relationships between knowledge from different dimensions,the problem of semantic relations missing in tranditional ontologies is solved.Then,a brand new way to describe survey rule in SSDO is proposed.3)Research on technical stuffs annotation and knowledge retrieval method.In the session of knowledge acquirement,aiming at problem ofartificial dependence and inefficiencywhile usinggeneral annotation methodtodeal with the annotation work of rule-based document(RBD),the rule description paradigm of the style "Condition Object + Condition Constraint-Action Object + Action Constraint" is proposed.Six rule knowledge templates are designed,and a knowledge acquisition method via regex parser and dependency parser is proposed.During annotation of SD,the annotation method for minimum survey space(MSS),general survey position and walkable path is designed.Meanwhile,in order to realize the information fusion of SD annotation and survey rule knowledge,the method to transform abstract structure into individuals is proposed.Finally,the main framework for technical stuff retrieval is analyzed,and the semantic retrieval progress is discussed in detail.4)Research on task extraction,task simplification and task set allocation in SSGP.Aiming at technical limitation of task assist by traditional knowledge assist method,and combing with the basic characteristics analyzation of the MSS,survey tasks are divided into several categories,and the rules to extract task from survey rule are designed.Compared with historical survey tasks,the object tasks are simplified by the influence of the four factors,including the same names,similar names,similar names in inner structures and distances in MSS,and the simplification algorithm is designed and verified.Task set allocation is the basis of task allocation,while dealing with the problem,the influences of types,complexities,survey positions of the task is concerned.Finally,genetic algorithm is used to describe the object function,and related experiment is carried out.5)Design and demonstration of SSGP prototype.Based on the above theories and studies,the functional hierarchy of SSGP prototype is proposed,and the system is developed.After that,the introductions and demonstrations of meta-ontology building,knowledge acquisition from RBD,SD annotation and task management are carried out by steps.As is shown in the result,the prototype follows the concept and hierarchy of SSGP,and meets knowledge assist demand in ship survey industry.