The transmission system is the connecting part of the power station and, distribution is capable of being forwarded to the load center. If there is a fault in the transmission line by interrupting the electricity supply to the load, then this will cause a loss for consumers. Therefore, another technique is needed to identify the fault in the electrical power distribution system accurately and quickly by reducing search time and speeding up the repair process. This study will present a method to identify fault by classifying and estimating the location of a fault in the 115 kV transmission system. This technique is performed by combining Discrete Wavelet Transformation (DWT) and Recurrent Neural Networks (RNNs) of Elman. DWT aimed at extracting information of transient signals for each phase current and zero sequence current during one cycle when the fault starts. Elman RNNs are classified to detect a fault in each phase and ground, while Elman RNNs are used to measure the location of the fault in the transmission line. Training and testing data be carried out for the simulation of short circuit fault under different fault resistance and varying starting angle. Short circuit fault applied in the transmission line to 115 kV bus LK to BK on 63km line lengths. The fault classification results obtained are the accuracy of 100%, and the estimated location of fault received the most significant average error value is 1.4%