In high voltage induction machines the stator slots usually are wide opened to facilitate the assembling of the stator winding coils. Thus the magnetically effective air gap and higher order harmonics are rising, the power factor is decreasing. To compensate this negative effect magnetic stator slot wedges are frequently applied. During operation these slot wedges can get loose and eventually fall out totally. Currently a detection of fallen out slot wedges is only possible by time consuming partially disassembling the machine and optical inspection. Simple and reliable testing methods can thus increase the reliability and reduce costs due to unnecessary disassembling of the machine. For such testing methods high frequency or transient electrical properties of an electrical machine suit very well as the base. When high frequency or transient voltage signals are applied to the terminals of the machine the resulting current response contains information about the machine's magnetic state. Therein superposed are the magnetic material properties, several inherent asymmetries such as spatial saturation or slotting, as well as fault induced asymmetries. This paper introduces a new signal processing chain to detect and isolate the fault induced asymmetries caused by fallen out stator slot wedges. The chain consists of data capturing by collecting current response values due to voltage pulses and following Fast Fourier transformations. Measurements for several slot wedge fault cases are presented. The measured and calculated results show the high sensitivity and reliability of the proposed method.

No abstract

No abstract

No abstract

No abstract