Study And Application Of Extension Association Rules In Alarm Correlation Analysis

Network shortens the distance in the world,an increasingly important role,the impact of network fault is becoming more and more serious.Traditional method of network fault diagnosis mainly rely on the experience of the staff for a long time to deal with the fault,this method is inefficient and training an experienced technical personnel to spend high cost,long training period.With the development of artificial intelligence,computer technology,data mining technology and the successful application of social life,data mining technology has become a hot research topic in the network fault alarm correlation analysis.Analysis of the characteristics of network alarm correlation and apply to the network fault location,can improve the efficiency of the fault location,has a high practical application value.The paper makes data mining technology and extension knowledge as the base,the telecommunication network fault alarm correlation analysis and fault point as purpose,then finishes the theoretical research on algorithm of extension association rule analysis of alarm correlation,and through synthetic data sets,using C# programming language to do simulation experiments,the data were analyzed by Matlab mapping,rationality to validate of the algorithm.Firstly,this paper analyses carefully an example of alarm correlation analysis based on Apriori algorithm,and analyzes the bottleneck of the application of this algorithm and its improved algorithm in the application of alarm correlation analysis.Then,according to the characteristics of network fault alarm data,with the extension knowledge,the extension matter-element model be used to expression the key attributes of alarm information formally,which is suitable for data mining algorithm and reduce the amount of data from the source;according to the properties of the alarm information characteristics and time characteristics,the vertical and level weighted called hybrid weighted algorithm be used to reduce the redundant rules and improve the efficiency of fault diagnosis.In order to reduce the influence of subjective factors in the process of setting the level of weight coefficient,according to the characteristics of the alarm instance attributes,each attribute value of the objective function considering the level of weight,using the correlation function to calculate the similarity between the alarm and determine the weight value.Finally,the method of real-time diagnosis of network fault is analyzed.According to the characteristics of extension association rules,the method of extension distance is used to realize the matching of alert correlation rules,so as to realize real-time fault diagnosis.