Application Of Rough Set Theory In The Fault Diagnosis Of Distribution Network

As the electric power develops rapidly and the power distribution network grows larger, the dangerouseffect of its malfunction on the power distribution network itself, the agriculture production as well as thedaily life also increased. The inevitability of malfunction of the power distribution network makes the faultdiagnosis rate one of the key indicators for the reliability of power distribution. Consequently, it is of greatsignificance to find an efficient way of fault diagnosis for maintaining the safe operation of powerdistribution network and improving the reliability of power distribution.The fault diagnosis of the power distribution network could thought to be a question of classification,the rough sets is based on classification and is of great capability of fault-tolerant. So, this article proposeda diagnostic method based on rough sets theory for the power distribution network in the case of completemalfunction information. It can get the five major categories of fault diagnosis rules for distributionnetwork are electric interval, transformers, lines, bus and power failure. Though the rough set theory is ofgreat capability of fault-tolerant, the lost or mutation of core property in the date system would make it amissing or mistake decision. In this case, we use RBF neural network in the MATLAB neural networktoolbox to handle the malfunction.But in real situation, as there are many uncertainties such as accident movement or immovability ofthe circuit breaker and information distortion, the breakdown information collected is often uncompletedand uncoordinated. At this point, processing problem with classical rough set theory and neural networkswill result in unsatisfactory or error. As the maximal consistent block technique (extension of classic roughset theory) is of great accuracy in the uncompleted information system, this article proposed a diagnosticmethod based on the maximal consistent block technique for the power distribution network firstly. Adecision table is established with the change action signal of protection and circuit breaker statusinformation as a condition attribute and possible fault conditions as a target attribute by using the rough settheory. Then, using the knowledge of the maximal consistent block technique and generalized decisionreduce attributes, and eliminate redundant attributes in order to get ten determine diagnostic rules and twooptimal generalized decision rules for the distribution network fault diagnosis, and it extracts optimaldecision rules which illustrate the correctness by the examples. It could not only improve the diagnosticrate, but also reduce the diagnosis time, and finally help the workers to recover operation of the non-faultregions and overhaul the fault equipment.As a Soft computing method and artificial intelligence method, rough set theory has a widespread usein lots of fields. This article used the rough set theory in the diagnosis of power distribution networkmalfunction, in the case of uncompleted or inharmonious fault information, maximal consistent blocktechnique solve this problem excellently. Using maximal consistent block technique gives a newtheoretical method in the diagnosis of power distribution network malfunction.