Application Of Multidimensional Multiple Fuzzy Reasoning Algorithm In Network Fault Diagnosis

In today’s society, human activities can not be separated from the support of the network. Therefore the quality of network services is particularly important. In order to avoid a significant impact on people’s lives, once the network failure, the network administrator must repair the network quickly and recover the normal operation of the network.The network node will send root alarm information when it breaks down. The root alarm information spreads in the network and causes other related network node also to send alarm information. Ultimately, a large number of alarm information related to the root alarm is generated in the network. In order to help the network administrator to quickly find out the root alarm and then to locate the failed node, it needs to remove redundant alarms by alarm correlation analysis. However, the ambiguous relationship between alarms brings huge challenges to establish the correlation model between the alarms. Applying fuzzy theory can solve this problem effectively. At first, use the fuzzy association rules data mining algorithm to establish the correlation model between the alarms. Based on this, use the fuzzy reasoning algorithm to analyze the correlation between alarms and remove redundant alarms. Finally, the purpose of quickly finding the root alarm information and locating the failed node will be achieved. This paper focuses on the following three points:⑴The multidimensional multiple fuzzy reasoning algorithm. In order to reflect the ambiguity of the alarms and the relationship between them, and also to enhance the fuzzy rules’ rationality and ability of representation and reasoning, this paper chooses the weighted multidimensional multiple fuzzy reasoning model for reasoning, and proposes a reductive weighted multidimensional multiple fuzzy reasoning algorithm.⑵How to obtain the weights in the weighted multidimensional multiple fuzzy reasoning model. It is very difficult to get these weights. In this paper, the rules of the reasoning model are mapped into a fuzzy neural network, and the weights in the reasoning model can be obtained by training the fuzzy neural network.⑶The control strategy of the fuzzy reasoning system. Because there are huge amounts of alarms in fuzzy alarm database and a large number of rules in fuzzy association rules database, to speed up the matching and searching of alarms and rules, this paper proposes a matching and searching strategy based on grouping and sorting.Simulation results show that applying fuzzy theory to network fault diagnosis and using fuzzy reasoning method to analyze the correlation between alarms can remove the redundant alarms and locate the failed network node quickly. Use the reductive weighted multidimensional multiple fuzzy reasoning algorithm proposed in this paper can reflect the fuzzy relationship between alarms better, make the result more accurate and improve the accuracy of network fault location effectively.