Research On Unified Modeling Technology Of Manufacturing Big Data Based On Domain Ontology

With the continuous integration of science and technology with manufacturing industry,such as Internet of Things,artificial intelligence,big data,etc.,the manufacturing industry has been developed rapidly.In addition,manufacturing big data is the core competitiveness of enterprises in the era of "Big Data" as a potential wealth.Therefore,the integration and management of manufacturing big data is especially necessary.At present,the data generated by the manufacturing industry in the production process has the characteristics of multiple sources and heterogeneity,which face many problems such as difficulties in data aggregation,lack of uniform data representation and difficulties in data integration.In recent years,with the development of unified modeling technologies for big data(RDF metamodeling,networked modeling,etc.)and data integration management platforms(EDM,PDM,Predix,etc.),which enable the manufacturing industry to improve the integrated management of manufacturing big data.But,the existing technologies and platforms mainly focused on specific scenarios,which are weak in unified integration and management capability of manufacturing big data in different scenarios,so that the enterprises lack enough effective data as the foundation support.In this paper,focusing on the characteristics of manufacturing big data with multiple sources and heterogeneity,a unified logical view model(Meta-Onto-M)based on domain ontology is proposed which is based on the research of unified modeling technology for manufacturing big data,that includes source data convergence model,meta-model,ontology model,and finally a dataspace system is developed.In this paper,the major research contents and contributions are described as follows.1.In order to converge heterogeneous manufacturing big data from multiple sources,this paper proposes the manufacturing big data convergence model(NAP),which firstly processes the internal data of the manufacturing system and the external data of the manufacturing system,and then organically converges the data involving the whole product life cycle process.The NAP model can effectively avoid the problem of missing and insufficient data in the system,thus improving data utilization.2.In order to represent the heterogeneous manufacturing big data from multiple sources in a unified format,this paper proposes a meta-modeling scheme for manufacturing big data,which categorizes "Man-Machine-Material-Method-Environment" data from coarse-grained to fine-grained for metadata characterization in hierarchical manner.Firstly,it designs the manufacturing big data metadata model from a bottom-up approach that is divided into instance layer,and then model layer and meta-model layer.Secondly,it builds a coarse-grained unified resource description for data,which includes basic data,extensible data,data path,and identification information.Finally,it designs the fine-grained metadata for the "Man-Machine-Material-Method-Environment" data.The Meta-modeling proposed in this paper can effectively integrate the source data,so that multiple sources of heterogeneous manufacturing big data can be represented in a unified structure.3.In order to integrate heterogeneous manufacturing big data from multiple sources,this paper proposes an ontology modeling scheme for manufacturing big data,which can be dynamically integrated through hierarchical and extendable triples.Firstly,it adopts the "seven-step approach" to design ontologies by top-down classification while combining with Dublin core to build domain ontologies.Secondly,Meta-Onto-M model is proposed in combination with meta-modeling,which is used for describing ontologies by OWL language and obtaining domain knowledge documents with OWL description.Finally,various types of data and their relationships are integrated based on Pay-as-you-go evolution and integration mechanism.The ontology modeling proposed in this paper enables the multi-source and heterogeneous manufacturing big data to be integrated with a loosely coupled way which breaks the time and space restraints and improves the query efficiency.4.In order to manage multi-source and heterogeneous manufacturing big data,this paper designs and implements the dataspace system based on B/S architecture.Through the process of source data convergence,meta-modeling,and ontology modeling,the domain ontology is described with OWL language in a unified manner,and finally domain knowledge is formed and then imported into Neo4 j graph data.This paper designs and develops a Changhong dataspace system based on the data generated during the actual production process of Sichuan Changhong,which achieves the unified integration and management of Changhong manufacturing big data.