Research On Key Technology Of Cloud Manufacturing-based Knowledge Service System Of Product Design And Its Application

As the rise of manufacturing informatization,globalization and collaboration,the construction and research of knowledge service system has become one of the focuses.The purpose of this paper is to build a knowledge service system for complex product design and manufacturing based on cloud manufacturing platform,in order to achieve rapid acquisition and reuse of knowledge in the process of complex product design and processing,and to standard and unify the technology of various knowledge processing in the platforms of CAx?PDM,etc,to meet the needs of deepening development of knowledge engineering technology in manufacturing industry.Aiming at the knowledge requirement in the process of product design in manufacturing enterprises,we've developed the knowledge service system of product design,and study the data mining algorithm in the manufacturing field,which provides support for the designers in the manufacturing industry to quickly discover and integrate useful knowledge in the data ocean.(1)Face the feature of product design knowledge,we've developed a five layer framework of the product design knowledge service system,based on the study of the typical framework of cloud manufacturing system and the framework of the classic information system.The system use the virtual equipment technology,to virtualized the manufacturing equipment,experimental equipment,simulation equipment and other physical equipment that embody the manufacturing capability,and provided those virtual equipmen as the sevices to the system users to access and use at anytime and anywhere.The system use the technology of application service encapsulation to encapsulate all kinds of design knowledge,design process and product design example as a service,and the processing equipment is opened to the platform user in the form of service,too.The system privileged users can invoke these services by simple page input or parameter design according to their own needs.(2)Considering the variousness of product design knowledge,the product design information is sorted and structured through knowledge granulation technology,and the structure and properties of knowledge particles are used to reflect the relation and restriction of knowledge.Aiming at the problem that there are a lot of irregular knowledge in product design information,this paper proposes a method of granulating irregular knowledge based on rough set theory.The characteristic of irregular knowledge is that the structure is complex and difficult to be expressed in a standardized structure model,but it plays an important role in the knowledge service of product design.For the processing of irregular knowledge,the paper decomposes the irregular knowledge into standard terminology unit,text description unit,graph /model with parameter unit,design / process / calculate step unit,and establish key words and index words to connect them to each other and relate to the whole knowledge system.In order to apply to the network operation environment and to provide services on various heterogeneous systems without hindrance,this paper uses the knowledge expression of RDF?XML and STEP?XML.RDF?XML is used to express the design knowledge of text class products,STEP?XML is used to express the design knowledge of digital model products applied to CAx software.(3)The text similarity calculation involves the research of knowledge language participle,text knowledge classification,knowledge semantic recognition and so on.It is the key technology of text knowledge discovery and knowledge acquisition.Text similarity computation can be divided into three levels: word similarity computation,sentence similarity computation and file similarity computation.Similarity computation methods can be divided into structure-path similarity computation,corpus based similarity computation and mixed similarity computation.This paper constructs the granularity structure and corpus of product design knowledge,providing the implementation basis and convenient condition for text similarity computation.In view of the structure and characteristics of process sequence knowledge of product design,this paper deals with the process sequence knowledge based on aggregation clustering analysis,and designs an improved method based on Apriori algorithm—“suq?apr?gen”,as a knowledge discovery algorithm.The core of suq?apr?gen algorithm is to adopt a frequent itemset algorithm based on association rules for sequential knowledge,and mine frequent itemsets through candidate set generation and downward closed detection.(4)In the process of complex product design,knowledge reuse is frequent.This paper presents a network structure that combines CNN and RNN and finally classifies through the full connection layer(softmax)to learn the internal characteristics and correlation of non-standard product design knowledge.Combined with the design material knowledge of a military product,the feature vector set and key network parameters of the learning model are given.(5)Knowledge reasoning is an important technical means in the knowledge service system,which can be divided into rule based reasoning,model based reasoning and case based reasoning.Case based reasoning is a kind of similar or analogous reasoning method.Its core is to find a similar source instance from the knowledge base,take out the solution from the source instance and take it as the starting point of the target instance,get a strategy of the target instance solution by proper modification,solve the problem with known examples or experience.This paper studies a strategy and method of variant reasoning based on rough set,and gives the reasoning design process,calculation method and design example.(6)Knowledge push is a actively knowledge service that knowledge according to the potential design requirements of designers.This paper analyzes the forerunner,complexity and characteristics of designer's design intent in the process of design and processing.Combining the knowledge structure characteristics,concept similarity and text similarity calculation,a design knowledge push method based on design intent is proposed.