Applied research of real time fault diagnosis technology for power system based on MAS

For improving diagnosis speed and accuracy of large scale and complex system like satellite, a distributed diagnosis frame of satellite power system is put forward based on MAS. The most important parts of the frame are two types of diagnosis agent. One is adaptive global agent and responsible for global diagnosis, and multi signal model is adopted to diagnosis; another is subsystem resident agent and responsible for local diagnosis, and parameter estimation diagnosis strategy is adopted. Box-Jenkins modeling method is introduced. Through communication agent and power expert agent, coordination operation between agents is realized. The results of applied to power system show that goodness of two types of diagnosis agents are complementary, the diagnosis efficiency is improved and diagnosis outcomes are accurate.