Development of Transformer Condition Monitoring Software by Frequency Response Analysis

At present, the major problem with transformer fault detection using the FRA is that all analysis methods are highly complex in terms of algorithms and computations that are not easily used by the exploiter. Unlike the DGA method, which has developed good soft-wares to interpret the results of its measurements, the FRA method lacks this possibility due to its fledgling. Therefore, it is necessary to develop accurate, user-friendly and reliable software for this purpose. In this paper, graphical software is developed in MATLAB environment to enable transformer operation engineer to monitor the transformer condition without performing complex calculations. For this purpose, first the necessary experiments are performed on different transformers (in healthy and different fault conditions) and then by past well-known methods, the type and severity of fault are determined. In addition to the previous methods, a new method based on the combination of probabilistic neural network and numerical indices for the detection of fault type is proposed. Also, a new criterion for detecting fault occurrence based on CC index (for case where transformer healthy condition measurement results are not available) is proposed. Finally, the performance of previous methods and proposed methods is evaluated and evaluated.