Neural Network Approach for Fault Location in Distribution Networks with Limited Measurements

this paper presents new method for faultlocation in radial distribution feeder in which artificial neural network and the measurements of the limited points along the radial distribution feeder are used to enhance the accuracy of the fault locator. Conventional fault locator uses recursive algorithm of load flow in order to estimate the loads in each tap of distribution feeder before fault. Same algorithm is used to determine the fault location based on collected data during fault span but load in radial distribution feeder changes unexpectedly. Because of this uncertainty of load behavior most of the time, higher degree of error is unavoidable between real amount of voltages and currents andcalculate one at the end of feeder. this fact affects the calculation of fault locator algorithm severely. The ANN approach proposed to decrease these differences by using limited measurement along distribution feeder. Test result would be very useful information if used by artificial neuralnetwork to decrease the error in the actualproject.

Keywords:

Distribution Automation , Fault Location , Feed Forward Neural Network and RadialBasis Probabilistic Neural Network