A new EMD-Shannon entropy-based methodology for detection of inter-turn faults in transformers

Today, transformers are basic elements and of great importance in electrical systems. Unfortunately, they are prone to failures, including short-circuit failures. Therefore, it is necessary to apply methodologies that help to detect faults in a timely manner to avoid damage to the transformer. In this paper, it is proposed a new methodology to identify short-circuit faults by means of a technique called Empirical Mode Decomposition (EMD), which decomposes the signals into new signals called Intrinsic Mode Functions (IMFs). These signals are analyzed with the entropy of Shannon in order to obtain indices that help to characterize the severity of the fault. The analyzed signals correspond to both a healthy condition and different short-circuited conditions, i.e., 10, 20, 30, and 40 turns. Results demonstrate the effectiveness of the proposal.