Research On Question Answering System Of Knowledge Base In Intelligent Management Of Electromagnetic Compatibility

With the development of modern science and technology,the number and types of electronic equipment are increasing,which makes the electromagnetic environment of electronic equipment increasingly complex.At the same time,the integration of electronic devices is getting higher and higher,and the anti-interference ability of electronic devices is declining.Therefore,it is very important to ensure that the electronic equipment can work normally in the complex electromagnetic environment and reduce the electromagnetic interference between them,which puts forward high requirements for the electromagnetic compatibility of mobile phones and other electronic products.So it is more and more important to consider EMC in mobile phone design.At the same time,with the continuous progress of artificial intelligence technology,more and more artificial intelligence technology is applied in actual production.Integrating artificial intelligence with EMC management to solve practical problems is a topic worthy of exploration and research.By using artificial intelligence technology,automatic EMC management can be realized and human and material resources can be saved.In this paper,an interactive question-and-answer system based on knowledge atlas is used to realize automatic EMC management.According to the user's problem description,combined with the characteristics of electromagnetic compatibility field,named entity recognition is carried out,the electromagnetic compatibility entity is extracted,the entity is linked,and then the existing entity in the corresponding knowledge map is obtained.Because of the special relationship between EMC entities,we designed our own special knowledge map(functional relationship triple).According to the relationship triple of knowledge map,we conducted interactive question and answer,got the definite relationship between entities,and returned the corresponding solution.In this paper,deep learning is used for named entity recognition.Convolutional Neural Networks(CNN),Recurrent Neural Network(RNN)and other different neural networks are tried.Through transfer learning,we can train more accurate expression of word vectors and improve the recognition accuracy.Because of the excellent performance of conditional random field algorithm(CRF)in sequential marking tasks,which can take into account the tag information before and after the sequence,we try to add CRF layer in the last layer of the model to improve the accuracy of sequential marking.In order to enhance the learning ability of the model,the morphological features of words and the vocabulary features of EMC specialty are also added.The model fusion technology is tried to improve the accuracy of final recognition.In this paper,we use interactive question and answer(IAQ)to obtain the attribute values of the functional relationship triple according to the interactive question and answer between users.Then,the relationship between entities is determined according to the functional relationship triple,and the corresponding EMC management scheme is returned according to the acquired attribute values.Experiments show that the system has a good performance in actual use scenarios,and the combination of artificial intelligence and EMC management is feasible.