Most numerical protection relays in Viet Nam use single ended fault location algorithms. While this method is simple and fast, the following commonly encountered factors can severely degrade accuracy such as high fault resistance, zero sequence, mutual coupling and non homogeneous power systems... This paper presents fault location method for overhead transmission lines based on Artificial neural network (ANN), then prediction model and simulation analysis of 110kV overhead line have built by Matlab software. The trained ANN for fault classification and fault distance location was used Levenberg Marquardt training algorithm. This method selects magnitude of current and voltage form all phases of the overhead line as the inputs of the ANN that can output fast and precision result. The result shows that the ANN combines as analysis tool to provide fault information that is more reliability and accurate than provided by a traditional electromechanical relay and it also overcomes the limitations of the single ended fault location algorithms.