Method for building network fault diagnosis rule base

Abstract

The invention relates to a method to build network fault diagnosis rule base, which comprises: (1) inputting fault data by network device; (2) pre-treating the condition property of data; (3) building up a decision information table from pre-treated fault data; (4) reducing the table; (5) producing and storing the rule base. This invention has least rule number and property, uses little storage space, and has faster diagnosis speed.

Description

technical field [0001] The invention relates to a computer network, in particular to a rule base for diagnosing computer network faults, in particular to a method for establishing a network fault diagnosis rule base. Background technique [0002] In computer networks, network management software generates a large amount of data every day and stores them in the network management database, including MIB variables obtained by polling (management information base variables that come with equipment), network performance measurement data, and network alarms. Network administrators are overwhelmed by a large amount of information, unable to discover the knowledge existing in the data, resulting in "data explosion and knowledge poverty". How to use knowledge discovery technology to mine novel knowledge from these data and perform fault prediction and fault automatic recovery is a very promising technology. Related technologies include network alarm database knowledge discovery bas...

AI Technical Summary

Problems solved by technology

However, this method has the following defects: 1. This method can only mine MIB variables, and cannot involve other types of network information processing (such as performance parameters, etc.); 2. The step of reducing the decision information table does not consider the cause of the failure The similarity between the rules, the number of rules, the attributes contained in the rules, the rule storage space is large, and the speed of fault diagnosis is slow; 3. The acquisition of network fault knowledge is limited to mining all previous fault data. In the new network management When data is added, knowledge acquisition operations can only be performed on new and old data together, which is inefficient

Images