Research On Fault Diagnosis Method Of Magnetic Suspension Bearings Integrating Knowledge And Data

Due to its compact and reliable structure,low calorific value,low power consumption,adjustable stiffness and damping,and automatic balance,the maglev bearing system has been widely used in military,aerospace,high-speed machine tools and medical equipment and other refined and high-end manufacturing equipment.Industry,which puts forward high requirements for the reliability,safety and stability of maglev bearings.Maglev bearing fault diagnosis has always been the focus of research in industry and academia.Since the maglev bearing is a system with complex knowledge,it is difficult to quickly find the root fault location and cause based on fault characterization alone,which increases the difficulty of fault diagnosis.In addition,the relevant knowledge and experience are not unified and integrated,which also affects the improvement of the fault diagnosis efficiency of the maglev bearings.Therefore,on the basis of comparing and analyzing the research status at home and abroad,this paper draws on the creation process and feature extraction method of knowledge graph,and combines the fault data of maglev bearings to study the fault diagnosis method of maglev bearings that integrates knowledge and data.The specific contents include the following four aspect:(1)Research on the creation method of fault map based on unstructured data of maglev bearings.By analyzing the characteristics of the maglev bearings fault text data,the data is preprocessed,and then an improved text vector generation method is proposed.A text semantic parsing framework is proposed and the string matching algorithm is improved to realize text content parsing,which lays the foundation for the creation of fault graph.(2)A variety of feature extraction and fusion methods of structured data are proposed to realize the classification of maglev bearing faults,and on this basis,a method for creating feature graph is proposed.By collecting and analyzing the structured data of maglev bearing faults,a feature extraction method of structured data is proposed.On this basis,the convolutional neural network is used to classify the faults.And the relationship extraction is realized through the fault results of the classification,and a feature graph creation method is proposed.(3)A fault diagnosis method of maglev bearing that integrates knowledge and data is proposed.Taking the fault graph as the knowledge engine and the feature graph as the data engine,the maglev bearing fault graph and feature graph are integrated through the graph fusion method.On this basis,the fault diagnosis experiment of maglev bearing with fusion fault and data is carried out,and the validity of the proposed method is verified by the interpretation accuracy index.(4)Development of fault diagnosis system for maglev bearing.Through the development of fault diagnosis software integrated with the proposed fault diagnosis method,the maglev bearing fault diagnosis is completed and a fault diagnosis report is generated by using the simple,convenient and efficient system operation.