Research Of Network Intelligent Malfunction Localization System

Fault management, in which localizing malfunction is the core task, is one of the five network management modules according to different functions. The localizing malfunction technique has been becoming one of research hot spots in the field of fault management.In this paper, both the rule-based method and the data mining method are employed in order to overcome the disadvantage resulted from traditionally using single method, which make the system have not only learning ability but also logical inferring ability. Therefore, the system includes two modules: rule engine module and mining rule module.As far as rule engine module is concerned, the great efficient RETE algorithm is implemented. In order to seamlessly integrate the rule engine with the application programmed in the object-oriented language, a native knowledge presentation method and Rete network are introduced. In addition, the traditional RETE algorithm is optimized.As far as mining rule module is concerned, based on an alarm event sequential model, the Robust_search algorithm is improved to avoid losing the sequences that are satisfied with our predefined needs.Both the rule engine module and the mining rule module, which are integrated, are implemented in java language respectively. The concrete designing process is illustrated in the UML graphs.