The purpose of the article is to provide potential tools that can make a significant contribution to the identification of partial discharges (PD). Different types of partial discharges occur in stator winding insulation and a few partial discharges may occur simultaneously. Internal partial discharges are electrical discharges that occur in voids in the insulation of the stator winding. In typical stator insulation systems that use epoxy bonded mica tapes, insulation degradation due to internal partial discharges is usually slow (many years or decades). External partial discharges (slot PD and surface PD in the end-winding) are more dangerous and lead to the destruction of the insulation in a short time (several months or years). Therefore, the identification of insulation defects is essential. The analysis of existing methods for identification of defects in the insulation of high-voltage electrical machines using the results of measuring the partial discharges characteristics is carried out. The advantages and disadvantages of each of the groups of identification methods are characterized. It is shown that among the models of knowledge representation in solving problems of diagnostics of insulation systems for high-voltage electrical machines, identification methods, including field tests using training samples, are among the most suitable ones. It is noted that detection of insulation defects and their identification cannot be carried out only by direct measurements of PD characteristics and other dielectric parameters (electrical resistance, dielectric loss, polarization index). For this, special computing programs based on pattern recognition methods should be used. Results are presented of identification of technological defects in the insulation of the stator winding at the stage of factory testing, obtained using the PD identification method developed by the authors
