| Nowadays, with rapid development of information technology and the universality of the internet technology, all kinds of information system is becoming more and more popular. There is a problem for the network management personnel to ensure that the complex network is able to run efficiently, safely, reliably and stably. The traditional network fault management system is hard to adapt the expanding scale and complex internet condition because of excessive reliance on network experts. We can make up for these shortcomings through using data mining and event correlation technique.This paper achieves good results by applying the data mining and the event correlation to network fault management and analysis system. The main research work is as follows:The paper researches the mining of complex network of frequent alarm sequential pattern in detail. Traditional sequence patterns which are based on the WINEPI algorithm require frequent repeated scanning data, as a result, the algorithm’s execution time and efficiency is greatly reduced. Based on the FP-growth algorithm, this paper designs a FSOFP alarm sequential pattern mining algorithm which improves the execution efficiency of frequent alarm sequence mining greatly.Based on FSOFP algorithm, according to the frequent changes in the actual situation of alarm database, this paper presents the SC-FSOFP algorithm based on the minimum support degree minSup changes. The experiment results show the algorithm performance and effectiveness.This paper designs and implements a complete network fault warning and analysis system model by applying the data mining and the event correlation technique to the network fault management and analysis system. And then, this article designs the event collection, data preprocessing, knowledge base, inference engine and database in detail. At last, the system is implemented by writing code. The system can support various association rules and the frequent sequential pattern mining. |
PDF Full Text Request