CNC Fault Diagnosis System Based On Knowledge Graph

With the continuous transformation and upgrade of manufacturing industry,modern manufacturing industry has entered the ear of Industry 4.0.A variety of new products and technologies have emerged continuously and the technologies between different fields has cross applied.The CNC(Computer numerical control)machine is a complex product.So the faults of CNC machine need to be located an repaired when facing with them.Therefore,it is necessary to develop a system of CNC fault diagnosis which using new technologies such as knowledge graphs and cloud services to implement the diagnosis of CNC machine tool equipment.An CNC fault diagnosis system based on knowledge graph is designed and implemented in this thesis.The pre-construction of knowledge graph for fault diagnosis,the design of diagnosis method based on knowledge graph,system architecture design,the specific design for each module in the system and test of the developed modules is completed.First of all,the entity recognition task was completed using deep learning technology,realizing the pre-construction of knowledge graph.Then,a similarity algorithm is employed for the relationship between the users.The dynamically updating weights are used for the relationship stored in knowledge graph.A search algorithm is proposed for the search function.All the three methods can be beneficial for the diagnosis.After that,the system is divided into three parts which includes back-end application,database and font-end web interface.The back-end uses java language and Spring framework to implement logical processing function in the system.Neo4j can be applied to store knowledge graph data and MySQL can be used to store other data in the database.The web application provides the users an operating interface.The system can be divided into user information module,case information module,fault diagnosis module,phenomenon&remedy maintenance module,question center module and others by function.At last,the reliability of system is verified by corresponding functional tests and performance tests.With the increasing of fault cases in knowledge graph and the use of the system by users,the system developed in this thesis will be more and more accurate in diagnosis results recommended to users.And fault diagnosis based on this method may also be applied in other kinds of facilities.