The Fault Diagnosis For Power Grid Based On Multi-agent System For Rough Set And Bayesian Network

As the scale of the power system network in China has kept expanding since the21st Century,the power grid fault diagnosis on the grid in large regions is extremely important. Thus, accuratefault diagnosis methods fully guarantee the smooth operation of power grid. Nevertheless, in theprocess of power grid fault, the misoperation of the circuit breaker and the information loss orbias during the transmission would lead to information uncertainties, which makes it difficultfor power grid fault diagnosis to obtain relatively complete diagnosis information. In order tomake an accurate diagnosis, the fault tolerance of diagnosis system has to be enhanced. Only inthis way, can we accurately locate some of the faults and reduce the effects caused by them.First, the essay gives an introduction of the background and significance of the power gridfault diagnosis research, along with relevant research both in China and out of China, and brieflyillustrates some common fault diagnosis technologies. Furthermore, it introduces Bayesiannetwork including Bayesian structure as well as its characteristics and independence, and induceits building ways. It explains the basic concepts of Rough Sets Theory ranging from knowledgedelivery of the Theory to the reduct of diagnostic table and the ideas on reduct and core, and theAgent composition with its system structure, features, communication and cooperationincluded.As the core of the essay, the last part offers Multi-Agent System fault diagnosis basedon Bayesian and Rough and gives the diagnosis procedure. Combined with a classical powergrid structure, the experiment has been conducted. The calculation for each phase along with theresults has been displayed. Through all the calculation results under different circumstances, themethod has been proved both correct and advantaged, and Multi-Agent System Diagram basedon Bayesian and Rough has also been designed including the function setting and coordinationprocess for each Agent unit.