Research on Non-destructive Intelligence Detection of Tightness of Stator Slot Wedges of Large Hydro-generator

Abstract

Abstract In this study, the classification threshold of stator slot wedge tightness is established, and a non-destructive testing technology of stator slot wedge tightness based on acoustics is proposed. Firstly, the classification threshold of slot wedge tightness is determined. Then through the compression test of corrugated plate, the corresponding corrugated plate shape variable is obtained. The mechanical classification threshold is transformed into the deformation threshold of corrugated plate. Secondly, the tightness test platform of stator slot wedge is established and the tightness test is carried out. Then, the percussion signals is preprocessed by endpoint detection method, which greatly reduces the subsequent calculation. Subsequently, the characteristic parameters of the percussion signals are extracted by time-domain and frequency-domain analysis, and the characteristic parameters are screened by F-ratio method and Pearson correlation coefficient method. Finally, the Support Vector Machine(SVM) optimized by cross validation method is used to classify the test sets to realize pattern recognition. The results show that the tightness of stator slot wedge can be accurately detected by using this research. This research realizes the intelligent identification of the tightness of the stator slot wedge of the generator set, and provides an important reference for the tightness detection of the stator slot wedge.