Research On Data Association Relationship Oriented To Data Spac

Industrial data has formed a huge scale with the process of smart manufacturing and enterprise data,due to the different data collection,processing and storage methods among manufacturing enterprise,which makes the massive data sources complex and different structures;industrial data presents structured,semi-structured and unstructured characteristics,which leads to the the isolation of information communication among enterprise systems;dataspace can effectively solve the problem of efficient management of multi-source heterogeneous data,so dataspace has become a hot spot for experts and scholars to study in the industrial.Data association technology is the key technology to build industrial dataspace,in recent years,a large number of researchers have adopted graph models and related methods to study the association relationships among data resources,compared with traditional relational models,semi-structured models,and hierarchical models,graph models have a simple structure,using a node to represent entities,and the association relationships between entities are represented by edge,which are more suitable for multi-source heterogeneous and spacedata descriptions,and analysis and research based on graph models have attracted high attention from industry and academia.This paper applies the industrial dataspace technology to the aluminum electrolysis industrial scenario,studies the association relationship of data resources in the aluminum electrolysis dataspace,constructs the overall architecture of the aluminum electrolysis dataspace,builds a BiLSTM-CRF model to process electrolysis text data resource entity extraction,and improves a connected branch clustering algorithm based on gray correlation weighting.The details are as follow.(1)Apply the idea of Knowledge mapping to study the association relationship between dataspace resource entities,build a Bi LSTM-CRF model to extract the data resouce entities of aluminum electrolysis,then study the association relationship between entities using the dependency syntax analysis method,define the concept of domain entities,and construct the association map of aluminum electrolysis entities using Protégé software according to the mechanistic association between each logical entity.(2)Define the transaction space D of the electrolyzer,define the set of transaction topological space attributes Ω;introduce gray system theory,combine the gray correlation with beaten distance to ontain a gray correlation-weighted Euclidean distance metric,and obtain a gray Correlation-weighted connected branch clustering algorithm the to correlate analysis the historical data of aluminum electrolysis process parameters.(3)Taking a large domestic electrolytic alumnum enterprise as the background,the overall architecture of aluminum electrolytic dataspace is constructed,including cross-source heterogeneous data layer,information model layer,connector layer and data exchange lager.Then the integration and management principle of aluminum electrolysis dataspace data is anaiyzed,and the industrial data integration and management system is divided into resource layer,logical layer and application layer,and finally the visualized intelligent search function is realized based on entity association mapping.