Time-Frequency Analysis for the Interpretation of FRA Measurements

Inhalt:
Frequency Response Analysis (FRA) is a powerful and non-destructive test to diagnose the internal condition of a power transformer. This technique enables the detection of different kinds of faults related to mechanical and electrical problems that occur inside the transformer without the need of an internal inspection. This feature aroused not only the interest of researchers in the field of transformer diagnosis in the last decade but also has captured the attention of transformer manufacturers and electrical utilities because of its potential to verify the internal integrity of transformers, detecting incipient faults as a preventive strategy of maintenance or confirming failures after catastrophic events like lightning, external short circuits, earthquakes. FRA works as a comparative method, i.e. the detection of a damage or abnormal condition is done by comparing a current set of FRA measurements versus a well-known reference signature or fingerprint, that is, a measurement performed on the transformer in sound condition. The main goal is detect differences between FRA traces which are helpful to infer the condition of the unit. The standardization of FRA tests is nowadays incomplete. There are technical reports, recommendations, standards and user manuals available for this purpose. These documents have in common some theoretical frameworks and definitions but unfortunately they do not give a consolidated methodology for the interpretation of FRA measurements. Today the industry and customers are seeking for more guidance on the interpretation of FRA results to avoid misinterpretations. Therefore, there is currently a necessity for the development of a FRA interpretation algorithm to be applied to condition assessment. In the last years some diagnostic interpretation methods based on equivalent circuits, parameter identification of transfer functions and statistical indicators have been proposed, but none of them has been entirely conclusive. This document proposes a new approach for FRA interpretation using a novel tool known as Time-Frequency Analysis (TFA) which has already shown successful results in other fields, such as detection of diseases or syndromes using bio-signals, speaker identification or detection of faults in loudspeakers using sound signals or machine condition monitoring and diagnosis using electrical or vibrational signals.