Grid-connected induction generator interturn fault analysis using a PCA-ANN-based algorithm

It is difficult for short circuits among the internal windings of one phase in electrical machines to be determined in a reliable and speedy manner. The failure currents that occur, especially in short circuits with a few windings, are at such low levels that they cannot be determined by relay systems. This results in growing faults and damage. In this study, we designed a model that can define winding failures successfully at very small levels by using the PCA and ANN algorithms. We tested the real-time faults and measured the system performance with the installed test rig. The developed protection model determined fault determination in very small (2.5%) winding failures with acceptable accuracy. The suggested model is a counter-speed, selective, flexible, and economical protection model that may be used for internal failures of electrical machines. It has a structure that may be used in different systems or kinds of failure with data receiving and software changes.