Fault detection identification for power transformers protection

Abstract

The stability of the power system is dependent on the reliability of the various components in the network. A power transformer is a critical component of a power system's transmission and distribution infrastructure. However, it is susceptible to a wide range of problems, which can lead to power outages, which in turn can have a significant economic and social impact. Detecting and analyzing internal faults in a power transformer is a complex process that requires the use of appropriate fault detection procedures to ensure that the related repercussions minimized. Internal and external defects can founded in transformers. In addition to asymmetrical faults and faults from line to ground and line to line, winding flaws and winding insulation failures can cause turn-to-turn or ground faults, depending on the kind of external fault. Magnetic inrush current, lightning strikes, long-term overload, and failure of the cooling system are all possible causes of insulation deterioration. Dissolved gas analysis (DGA) is frequently used to discover transformer faults in the early stages. With the Duval Triangle as a focal point, this thesis provides the basics of introduction to DGA transformer interpretation. Precision in DGA laboratory findings may impact the accuracy of DGA diagnosis, as demonstrated by this study Both the previous gas levels and the lowest gas levels in service above which diagnostics can be attempted are listed below. There are certain users who are apprehensive about using triangular coordinates even though the Duval Triangle approach the specified in the IEC Standard and through these public assessments. In addition to this effort, MATLAB that used to construct a specialized machine learning approach (MLT) for the detection and classification of transformer faults. Measurements on two separate sets of transformers, one in good working order and the other with various faults, provide the necessary data for MLT training and testing (axial displacement, radial deformation, disc space variation, and short circuit of winding). The suggested method for fault detection the projected to produce comparable results to existing approaches because of its excellent MLT facilities during the learning and testing stages.