Currently Electrical power system is represented as the complex artificial system throughout the world, as the economic and social advancements based upon consistent, intact, stable and economic models. Because of varied arbitrary causes, unintentional failures happen in electrical power systems. Based on this problem, this work aims to develop EDCNN to detect and classify the fault signals like sag transient and swell in the transmission line. Moreover, the wavelet decomposed fault signals are extracted and identify the fault on the basis of the decomposed signal using the EDCNN approach. Moreover, this paper presents the performance evaluation by deciding the Type I and Type II measures and RMSE measures. In the performance analysis, it evaluates the performance of EDCNN method to several existing methods. The experimentation outcomes examine that the proposed EDCNN method efficiently and classifies recognizes the fault signal in the power distribution system while comparing with the conventional methods such as linear-Support Vector Machine (SVM), quadratic-SVM, and Radial Basis Function (RBF)-NN and Gradient Descent (GD)-NN models.