download_pdf_ijert
Authors :- Egeolu Ifeanyichukwu , Onah John

Volume/Issue :- Volume 3 Issue 1

Google Scholar :-
 https://goo.gl/DF9R4u

Scribd :- 
https://goo.gl/1rC3Ve

Thomson Reuters :- https://goo.gl/3bkzwv

This paper proposes an algorithm for detecting, classifying and locating single phase to ground faults on electric power 415 volts distribution lines. Feedforward artificial neural networks have been employed along with backpropagation algorithm for each of the three steps in the fault location process which are fault detection, fault classification and fault location. To validate the proposed algorithm, the Michael Okpara University of Agriculture Umudike plant house to new female hostel415 volts distribution line is modelled using Power System Computer Aided Design power systems analysis tool.Simulation results have demonstrated that the fault location method has high accuracy and good robustness. After the test set has been fed into the neural network and the results obtained, it was noted that the efficiency of the neural network in terms of its ability to detect the occurrence of a fault was near precision. The confusion matrices show that the chosen neural network has 100 percent accuracy in fault detection. The artificial neural network chosen for fault detection, fault classification and fault location satisfies the mean square errorgoal of 0.001 by approximately 100 percent. The overall correlation coefficient of the various phases of training, validation and testing for the artificial neural network chosen for fault detection, fault classification and fault location is averagely 99 percent which indicates that the neural network target is able to track the variations in the neural networks outputs very well. The gradient and validation performance plots shows a steady decrease in the gradient and the number of validation fails is zero which indicates smooth and efficient training. This further implies that the neural network can generalize new data fed into it more effectively. The test phase performance shows that the average percentage error obtained for the neural network chosen for fault location for the single phase to ground faults is below 0.5 percent which is very satisfactory and thus the neural network can be used for the purpose of single phase to ground fault location.
Keywords:-Distribution lines, Fault Location, Artificial Neural Network, Single Phase to Ground Fault, PSCAD, MATLAB, ANN, MSE.s.