Data-driven Knowledge Graph Construction And Predictive Maintenance Of Machine Tool Fault Diagnosis

Nowadays the production workshop is gradually developing in the direction of digitization,networking and intelligence.CNC machines are a core element of production and reduce machine downtime,increase system reliability and extend service life problems that urgently need to be addressed.From a knowledge-technical point of view,the expression of knowledge and the analysis of the error diagnosis of machine tools offers a new idea for solving the problem of CNC machine tool failures.Based on the machine condition properties and the fault diagnosis,this paper build the structure of a system for creating a knowledge graph and predictive maintenance for diagnosing machine malfunctions,cleaning,extracting and filling in missing values from characteristic data about the condition of the machine tool and the diagnosis of machine faults.Key technologies were analyzed the creation of a knowledge diagram and a method for predicting and maintaining machine tool failures based on the merge of the domain knowledge diagram.The specific content of the study was as follows:First of all,according to the structural characteristics of CNC machine tools and the current requirements of enterprises for data acquisition of CNC machine tool characteristic state,the data acquisition methods of existing CNC machine tools were compared and analyzed.According to state data characteristics of CNC machine tools,the edge side calculation scheme was designed,and the ETL data cleaning and extraction scheme was proposed.In view of the possible data loss in the process of feature state data acquisition and transmission.In this paper,a method based on missing value of class center was proposed,and a fault case information table of CNC machine tool was designed to provide guarantee for subsequent knowledge mapping construction.Secondly,in order to realize the construction of machine tool fault knowledge graph,according to the lack of historical fault cases in the workshop,this paper obtains sufficient case sources through the web crawler,and establishes the ontology model of CNC machine tool fault domain;machine fault treatment and synthesis,identifies the entity of CNC machine tool fault domain based on collaborative training algorithm,and extracts the artificial abstract relationship.The cosine similarity was used for knowledge fusion,and finally the machine tool fault domain knowledge graph and its visual expression were constructed.Then,taking the characteristic state data of CNC machine tool as the data source,the predictive maintenance model of machine tool fault fusion domain knowledge graph was proposed.According to error evaluation system of CNC machine tools based on predictive maintenance,rough set data reduction was carried out on the characteristic state data to reduce the input of prediction model,in order to improve the working time of the numerical control model,and the sample was used.According to the prediction results and the predictive maintenance evaluation system,the maintenance scheme of CNC machine tools was recommended to extend the service life.Finally,a corresponding prototype system was developed based on the above-mentioned theory and technology.Three functional modules including migration calculation processing application module,machine tool fault diagnosis knowledge graph construction and application module,machine tool prediction and maintenance application were constructed,and the operation process of the developed prototype system was tested through cases,and the research methods and models were verified feasibility.