Sun Zhaohui,Zhang Deyun,Li Qinghai
(School of Electronics and Information Engineering,Xi'anˇˇJiaotongˇˇUniversity,Xi'an 710049,China)
Abstract:A novel automated fault management system based on distributed intelligent propagation(AFMDIP)is presented by using the techniques of artificial intelligence and distributed network management. The cooperative agents are introduced to improve the ability of system’s data collection. Neural networks and temporary cases database are applied by agents for alarm collection, classification and cases management. The agents first diagnose the faults in a domain, and the unsolved problems would be submitted to the management center that will-diagnose the fault based on the case based reasoning after analyzing all the information from each domain, and then transmit this case to the agents. Thereafter, the consequent similar faults will be diagnosed directly by the agents, so the efficiency of the fault management is increased. The information and cases sharing can be achieved. The experiment shows that the AFMDIP efficiently reduces the bandwidth consume in the management center and speed up the fault diagnosis process under large-scale and multi-faults situations.
Keywords:fault management;neural networks;case based reasoning;agent