Alarm Correlation Analysis And Prediction Based On Big Data Machine Learning

Due to the speed of networks' development and the popularity of mobile devices,people have put forward higher requirements for the stability of mobile communication networks.Stuffs of mobile network therefore must fix faults accurately and quickly.Although the alarm data generated by the network device can be used to ensure troubleshooting,a series of alarms will be generated due to only one device failure.In the mean time,the system formation logs by put the records such as operation records,abnormal information,and software running conditions into a collection.The communication network generates a large amount of log data every day,making it difficult to query and operate the log.In order to solve the above problems,this paper uses data mining method and machine-aided work to conduct an in-depth research on log analysis and the location and prediction of fault alarms.The main research work of this paper is as follows:1.This paper predicts the occurrence of specific alarm types from a point on network alarms,and proposes an association analysis method based on FP-Growth algorithm for communication network alarm information.The correlation between fault alarms is analyzed on the basis of several data sets by focusing on the alarms which have high correlation with specific alarms,to obtain the warning alarms and achieve the prediction and recognition for specific types of fault alarms.2.This paper analyses the relationship between log and fault alarm,and proposes a fault diagnosis and early warning method based on communication network log data.Firstly,the log data is filtered according to the alarm times.Then,different machine learning algorithms are used to analyze and process a large number of log data.Finally,the trained classifier is used to classify the real-time log datas,thereby realizing the prediction of network device fault alarm.3.Based on the above theory and simulation results,this paper designs an intelligent analysis system of log and alarm.Firstly,it designs the overall framework of the system,then focuses on the introduction of various functional layers and workflow,and finally designs the specific application interface.In this paper,data mining method is used to study and analyze the massive alarm and log data generated by communication network,and the validity and feasibility of using machine-aided network fault management is demonstrated.