Knowledge Graph Construction For Product Quality Analysis Of Machining Lines

With the continuous development and progress of the industry,the machining line of mechanical products has realized automation and informatization.Every step of the line processing/testing process will produce data,and if the data cannot be used effectively,it will not only waste the data,but also consume the storage resources.The storage,query and analysis of quality indexes such as pass rate are undoubtedly the key issues that people pay attention to in the process of industrial production.The knowledge graph for quality analysis of machining line products can process and store the generated data,conduct quality analysis,and obtain information such as product qualification rate and key stations.It can also query the data in an intuitive and convenient way.To construct such a knowledge map,it is necessary to extract entities and attributes according to the characteristics of the processing line,and analyze the data to extract the hidden attributes of the data.In the data processing stage,one-hot coding technology and FTRL algorithm are introduced to preprocess the data,which plays a role of data dimensionality reduction.The following quality analysis introduces CATBoost algorithm to conduct knowledge mining on the pre-processed data,analyzes the quality information(qualification rate)of the workpiece,and verifies the effect of FTRL algorithm combined with CATBoost algorithm by comparing different methods.At the same time,according to the different characteristics of the processing of the contribution to the final quality of the workpiece,the key station is analyzed,and based on the above analysis,the attributes of whether the work is qualified and whether it is a key station are extracted.Based on the knowledge extracted in the first two steps and the professional knowledge of the processing line,the entity,relationship and attribute are defined in detail,and the knowledge graph of the processing line is constructed.The Neo4 j database is used to store and query data,and the operations of adding,deleting,querying and modifying are supported to facilitate the management of the knowledge map of the processing line.Finally,based on the actual machining line data,a knowledge graph with the functions of storing,querying and obtaining the quality of the workpiece is established.