Marine Diesel Engine Key Parts Manufacturability Evaluation Research On The Key Technology Of Knowledge Graph Platform

In the era of widespread utilization of cutting-edge 3D digital definition technology,the manufacturing industry is compelled to adopt a highly adaptable product development model to meet the ever-evolving demands of the dynamic and sustainable market.Within this context,the critical components of marine diesel engines,serving as pivotal elements of the ship power system,are characterized by intricate structural features and convoluted marking information,posing challenges in terms of design and manufacturability evaluation.Moreover,the lack of proper storage and utilization of historical design information and process documents further restricts the exploration of comprehensive life-cycle knowledge pertaining to the historical model.To address these issues and enhance the design efficiency and manufacturability of such components,this study leverages the Model-Based Definition(MBD)model as the fundamental data model for marine diesel engine key parts.Subsequently,research is conducted on product manufacturability evaluation and knowledge graph construction technology,culminating in the development of a knowledge graph platform for evaluating the manufacturability of key parts in marine diesel engines.The ultimate aim of this endeavor is to identify design flaws at the early design stage and enhance production efficiency accordingly.Our main research elements are as follows:(1)The knowledge graph platform for assessing the manufacturability of key components in marine diesel engines was designed with a comprehensive technical approach.The research context of this study was meticulously analyzed,system function modules were delineated,and the platform framework was meticulously planned to cater to the specific requirements of manufacturing key parts in marine diesel engines.As a result,the overall technical route for evaluating the manufacturability of key components in marine diesel engines and constructing knowledge graph was successfully formulated.(2)A novel strategy for evaluating the manufacturability of critical components in marine diesel engines is proposed.To begin with,a convolutional neural network model enriched with a self-attentive mechanism is utilized to construct a model feature classifier,which serves as a data source for intelligent evaluation.Subsequently,the feasibility of the proposed method is verified through examples,and notable reductions in the influence of human factors on evaluation results are successfully achieved.(3)Studying a design-oriented knowledge graph construction strategy.We address the problem of decentralized storage of design information by proposing a new Machine ALBERT pre-training model,which is used as the semantic encoding layer of text,with Bi-LSTM as the label prediction layer and CRF added as the overall label optimization layer to build the entity recognition model.For the relationship extraction model,we add a shared character encoding layer and a CNN layer to classify textual relations,transforming unstructured information into structured triples,which are then stored in the graph database after entity alignment to lay the foundation for subsequent graph use.(4)An intelligent question and answer strategy based on semantic parsing is proposed.A natural question template matching method based on a plain Bayesian classifier is proposed to address the problem of low generality of knowledge graphs.The method combines a stuttering word splitting tool for semantic parsing of interrogative sentences and constructs database query statements to provide various functions such as query and knowledge quiz to improve product design efficiency by combining with front-end technology.(5)A knowledge graph platform for evaluating the manufacturability of key components of marine diesel engines was developed.Based on NX software secondary development technology,a marine diesel engine key parts manufacturability evaluation module and a knowledge graph module were developed on the basis of the above research.The practicality and effectiveness of the platform were verified by using the key components of diesel engines as validation objects.