No abstract

No abstract

No abstract

The drying process is one of the most important tasks in transformers manufacturing stages. The drying quality in this stage is directly proportional to transformer lifetime. In this contribution, drying process at manufacturing site has been evaluated by use of Frequency Response Analysis (FRA). The measured transfer functions of power transformer during the stage of drying are analyzed. Using Artificial Neural Network (ANN), a method has been proposed to give an estimate for required time for drying process. Results show that the ANN could well forecast the required time for drying if the ANN is trained using the measured patterns. The estimation obtained from this method is valid for all the transformers which have the same design.

This paper uses detailed model for computation of sectional winding transfer functions that are necessary for partial discharge (PD) localization in transformer windings. In order to increase the accuracy of PD localization in practical cases, the circuit of PD detection impedance is included in the model. Also to increase the reliability, the use of calibration pulses is included in the localization algorithm. By that, it is possible to optimize model parameters before employing them to find the location of discharges. The accuracy of the method is analyzed by measurements on a special prepared transformer in laboratory.