Discrimination of Inrush and Fault Currents in Protection of Transformer using ANN-DWT Technique

Author Name(s): Dr. Vinay Barhate, Parikshit Bangade, Dr. Sangita Deshmukh, Sudarshan Khond
Author Email: barhatevt@rknec.edu


The main objective of relay is to reduce the occurrence and duration of undesirable outages concerned to the power transformer. This can be achieved by avoiding false tripping and achieving a high operating speed relay. In order to achieve these objectives digital relays are employed which are based on various topologies and concepts, in this paper digital relay based on Discrete Wavelet Transform (DWT) and Artificial Neural Network (ANN) are presented. Often digital relay based on Fourier transform were prone to false tripping as they were unable to distinguish between inrush and internal fault current because inrush current is rich in second harmonic component and second harmonic component may be high enough under internal fault condition also. Even Some power transformers, with new improved core material designs for lower core losses but also older units under some conditions, produce low levels of second harmonics which makes it difficult for relays to operate correctly. Here the wavelet transform is used for signal processing and to extract features from transformer current signals in both time and frequency domain. The current of the transformer is decomposed into a series of detailed coefficients by using DWT. This extracted feature from DWT is used as a training data set for ANN, which will then take the trip signal decision. Four different conditions are analyzed (viz normal current, inrush current, LLL internal fault, and external fault current). Primary current is taken into consideration for DWT analysis in MATLAB. The tested model in MATLAB is then implemented practically by using python programming for DWT and ANN on raspberry pi3 model. Output of ANN algorithm is communicated to circuit breaker for executing “Trip” or No Trip” signal.


Unwanted outage of power transformer results in discontinuity in power supply for consumers which results in poor performance of whole power system. Protection of large power transformer is a very challenging problem in power systems for uninterrupted power supply. Conventional deferential protection scheme for power transformer uses second harmonic restrain principle for discrimination of internal fault and inrush current with the help of FFT (Fast Fourier Transform). Second harmonic restrain principle which is the ratio of second harmonic altitude to fundamental harmonics and this ratio is used for judging whether the current is inrush of internal fault, But well-known Fourier transform suffers some disadvantages which are given below: – i. It cannot detect the harmonic which are not integer multiples of the fundamental frequency (50 Hz). ii. Fourier Transform accounts for frequency analysis only for a given signal. iii. It assumes periodic and stationary signals but Inrush and Fault current are non-periodic and non-stationary. Above all disadvantages of Fourier transform are mitigated by wavelet analysis. Wavelet transform is used for feature extraction from the current waveform of power transformer (differential current). As Inrush current is neither stationary nor periodic in nature, hence DWT analysis is suitable for feature extraction of the current signals.


The proposed simulation work with ANN block is working exactly as it was desired to trip circuit breaker only on occurrence of internal fault. In hardware part the second code, which is developed for training ANN model is working properly, as it was able to achieve the accuracy up to 85% to 95% and loss is also minimized up to a satisfactory level in practical environment. The Third Code, which is developed program for prediction or recognition is working satisfactorily and it is observed that while recognizing whether to give “Trip” signal or “No Ttrip” signal to GPIO pins of raspberry pi3, it gives trip signal where it was intended to give trip signal

826 total views, no views today

Download PDF File

About the author: admin