Identification Of Fault Types Of Mobile Communication Base Station

The mobile communication base station is the main equipment of the mobile communication network,and its stable and reliable work is essential for the stable and reliable operation of the mobile communication base station.However,in the actual work process,base station failures occur from time to time,and when base station failures occur,communication operators can only receive a small amount of information: the time when the failure occurred and the time when the accompanying alarm occurred.Therefore,how to effectively identify the type of base station fault by using these small amount of known information,help the operation and maintenance personnel to quickly locate and repair the base station fault,so that the base station can quickly return to normal operation state,is a problem that communication operators have been concerned about.As a reflection of the abnormal operation of the base station,the alarm does not mean that the fault must occur;but once the fault occurs,it will be accompanied by the alarm.The correlation between the two is an important basis for us to identify the fault type of base station.However,there are two difficulties in using this relationship to identify the type of base station fault: one is that there are a lot of redundant alarms in the alarm information;the other is that the relationship between the fault and the alarm is uncertain,that is,the occurrence of one fault will cause a variety of alarms,and the generation of one alarm may be caused by different types of faults.In view of the above difficulties,this paper chooses the Bayesian network which has great advantages in solving the uncertainty problems as the core algorithm,establishes the Bayesian network model of base station fault type identification,and optimizes the model by removing redundant alarms.Through testing,the F1 score of establishing Bayesian network model to identify the fault type of base station can reach 0.912.The main work of this paper is as follows:(1)Analysis of alarm data and fault data of base station.Analyze the alarm data and fault data of the base station respectively,and analyze the correlation between the number of alarms and the number of faults of the base station,so as to provide support for the follow-up research work.(2)Mining the relationship between base station alarms and between faults and alarms.According to the characteristics of base station alarm data and fault data,an alarm transaction extraction method based on clustering and variable sliding time window and a fault-alarm transaction extraction method based on clustering and outlier detection are proposed respectively,then the relationship between base station alarms and the relationship between faults and alarms is mined.(3)Establishment of Bayesian network model for base station fault type identification.According to the relationship between base station alarms and the relationship between faults and alarms,a Bayesian network model for base station fault type identification is established,and the model is optimized by removing redundant alarms in Bayesian network to reduce the complexity of the model.(4)Evaluation of Bayesian network model for base station fault type identification.The real base station fault record data are used to test the Bayesian network model,and the results of the model test are analyzed.