New method for the quantification and classification of generator stator winding

HE Liang, MA Lixin, LIU Yukuan

(Department of Electrical Engineering School of OpticalElectrical and Computer Engineering, University
of Shanghai for Science &Technology, Shanghai 200093, China)

Abstract: Aiming at the problem of difficulty to quantitatively analyze the partial discharge at the early stage of the generator stator winding, and to determine the maintenance standard based on the level of the discharge, the partial discharge intensity, time and environment of stator windings and so on were researched. A method to install UV sensor in generator stator slot wedge was proposed. A new type of UV online discharge monitoring system was constructed.The method could effectively detect the partial discharge of stator windings by detecting the ultraviolet generated by partial discharge. Meanwhile, aiming at the degree of partial discharge, projection pursuit grade model was established and gray code accelerating genetic algorithm was used to optimize the projection vector to achieve quantitative classification processing. Discharge characteristic was extracted by using wavelet packet decomposition. The method was tested by an example, and the simulation was carried out with MATLAB software. The results indicate that the model is feasible and effective, and can achieve realtime detection of partial discharge of the stator windings and to quantify the degree of discharge, which is helpful for the power staff to solve security risks before the failure happened the extension of the generator life and has important practical application value for partial discharge fault detection.

Key words: stator wingding; partial discharge; UV detection; wovelet packet decomposition; projection pursuit