The algorithm forms the network describing matrix D on basis of the connection of the switches and the power flow directions between two adjacent switches,then the fault judgement matrix D p is formed by adding the field terminal unit(FTU)information about the fault overcurrents and their directions to the diagonal of the former describi.

It forms a network description matrix of the fault area on the basis of fault information which is from different work model,changes the description matrix and obtains a fault judgment matrix,finds out the concrete point that is fault according to the fault judge matrix.

An artificial neural network for fault distinguishing by sampling halve is presented,and the pretreatment data are amplified.

Furthermore,a fault matrix deduced from that model can resolve the problem that the numbers of measurable parameters are not enough to acquire the performance parameter.