Feature extraction using EMD and classification through Probabilistic Neural Network for fault diagnosis of transmission line

In the presented work, an efficient model for classification of fault of a transmission line is proposed. The scheme proposed is the combination of Empirical Mode Decomposition and Probabilistic Neural Network for the classification of ten types of shunt fault. Post fault current signals are used for feature extraction for further study. Empirical Mode Decomposition (EMD) method is used to disintegrate the post fault current signals into Intrinsic Mode Functions (IMFs) and excerpt assertive features from it. The characteristic features of the first nine IMFs of each phase are used as input variable in a Probabilistic Neural Network (PNN) based intelligent fault classification model. For evaluating the expediency of the technique proposed, it is evaluated on a 500kV, 300 km length transmission line for all the ten types of faults using MATLAB/SIMULINK. A broad range of simulation quality is taken to train and test the data. Results simulated suggest that the proposed approach is quick learner, robust and has higher classifying accuracy.