Design And Implementation Of Motor Fault Knowledge Crowdsourcing Acquisition System Based On BERT Model

With the development of the times,there are more and more types of generators for different scenarios.Including hydroelectric generators,turbine generators,etc.,there will be many kinds of generators.The maintenance and diagnosis of the various motors mentioned above is a current problem.Only experts in related fields can grasp their corresponding motors for the support of many engines.This knowledge is fragmented.There are no relevant specifications and industry standards,there is no certification in related fields,it cannot be used frequently,and it will even cause severe interference to the sorting and diagnosis of generator fault knowledge.According to the above requirements,a motor fault knowledge base needs to be made for experts and users in related fields.This system mainly uses two methods as the source of data collection.The first one is to distribute questionnaires to experts and collect and add experts’ answers to the motor fault knowledge base.Another way is to extract the core information in the existing articles to obtain relevant motor fault knowledge and store it in the fault knowledge base after confirmation by the user.Data preprocessing,model training,knowledge extraction,and ontology alignment are the key technologies.The core idea of these methods is to make fragmented knowledge structured and on a specific scale.Based on the above requirements,the currently accessible Internet technology.The sources of the data are from experts and the journal "Big Motor Technology," and answer the questionnaire of experts while extracting the knowledge about motor faults in the journal as the data source.The model training part uses the latest best model and uses its named entity recognition task to extract relevant descriptions of motor faults in the paper.The final results of extraction and collection will be partly noisy due to the different expressions and accuracy of the model.These noises need to be processed twice,including body alignment,expert confirmation,etc.After approval,this knowledge is stored in the fault knowledge base.This topic mainly focuses on combining the knowledge of motor faults with related natural language processing and knowledge graph related technologies.Improve the current status of motor fault maintenance.Simultaneously,the construction of a motor fault knowledge base can significantly help the development of related industries and reduce labor consumption.At the same time,it can deal with associated faults for the first time.