Prediction of winding faults between non-stationary signals: Granger causality analysis for lightning impulse testing of transformers

One of the important research areas in transient phenomena is sensitive recognition of fault in the complicated insulation system of transformer to make a decision about either go (acceptance) or no-go criterion after impulse test at factory. The existing methods are, in principle, able to indicate the delay in coupling between the signals, but inferring causality from the time delay is not always straightforward. In this paper, the time delay between simultaneously measured signals is calculated and further extended to identify the impulse fault in transformers using Granger causality analysis. The principle of Bayesian information criterion is used in Granger causality analysis to estimate the minimum number of lags. Granger causality analysis enables the calculation of neutral current signal for any input excitation of full wave and chopped wave impulse for typical power transformers. The experimental studies are performed on 60/90MVA (132/33kV) and 150/200/250MVA (500/275/33kV) transformers to prove the feasibility of the analysis. Copyright (c) 2015 John Wiley & Sons, Ltd.