Research And Implementation Of Communication Failure Ee System Based On Deep Learning

With the continuous development of automation and artificial intelligence technology,the management of fault events in the communication field is gradually becoming more efficient and accurate.For traditional communication companies,a centralized fault management system is usually used to handle fault events.In the fault management and control process,it is necessary to manually analyze the fault events according to the preprocessing rules and requirements,and provide auxiliary fault information,which is relatively labor-intensive.Therefore,with the help of deep learning technology,it is of practical value to apply the event extraction model to the process of communication operation and maintenance,automatically extract and sort out communication fault events,improve fault handling efficiency,and reduce operating costs.The main work of this thesis is as follows:(1)Research on the event extraction model in the communication field,the main work is divided into two parts,one is to propose a communication entity masking(Communication Entities Masking,CEM)strategy in the process of domain adaptation and task adaptation on the communication corpus,established a communication entity lexicon,and trained a pretrained language model CRoBERTa-CEM in the communication field.The second is to perform event extraction tasks on the public communication failure event case data set,and combined CRoBERTa-CEM to form a pipeline event extraction model CRoBERTa-CEM-ECAE(Event Classification Augment Extraction,ECAE)that first classifies events and then extracted trigger and argument words.(2)The communication failure event extraction system was realized.According to the specification of software engineering and the actual needs of the communication company’s failure management process,this thesis completed the system requirement analysis,outline design,function module design,database design and detailed design.The main contribution of this thesis is to construct a failure event extraction model suitable for the communication field,and conduct benchmark model comparison experiments and ablation experiments to verify the effectiveness of the model and improve the extraction effect of communication failure events.Then,a communication fault event extraction system was built to realize the extraction and whole-process management of communication fault events.It replaced the steps that traditional communication companies need to manually extract fault events in the fault management process and provided a complete confirmation and labeling mechanism,reduced manpower consumption and downtime,provided some support for communication fault handling.