Research On Non-destructive Detection Of Tightness Of Stator-slot Wedges Of Large Hydro-generator Based On Percussion Signal | | Posted on:2022-05-05 | Degree:Master | Type:Thesis | | Country:China | Candidate:W D Chen | Full Text:PDF | | GTID:2492306731479454 | Subject:Mechanical engineering | | Abstract/Summary: | PDF Full Text Request | | Electricity,as one of the pillars of the world’s energy security system,is extremely important to the development of human society.Large-scale hydro-generators are an important part of the power system,and their fault maintenance is related to the safety of the national power grid.The tightness detection and re-tightening of stator-slot wedges is an important part of generator maintenance.The stator-slot wedge is a structure used to fix the stator bar in the stator slot.When the slot wedge is loose,the stator bar vibrates under the action of alternating electromagnetic force.This will cause safety accidents in generator operation.The research object of this paper is the statorslot wedges of the generator of Gezhouba Power Station.The purpose is to design a non-destructive detection technology for the tightness of the stator-slot wedge based on acoustics.The specific work is as follows:(1)The classification standard for the tightness of the stator-slot wedges has been established.The electromagnetic force that causes the stator bar to jump is studied.The critical electromagnetic force of the generator for normal power generation and shortcircuit is calculated.The mechanical classification standard for the tightness of statorslot wedges is established.The tightness of the stator-slot wedges is divided into tight,slightly tight,and loose.These respectively represent that the stator bar does not be bounced under short-circuit and non-short-circuit conditions,and the stator bar does not be bounced only under non-short-circuit conditions,and the generator cannot work.Then through the compression experiment of the corrugated board,the corresponding corrugated board shape variable is obtained.The mechanical classification standard is transformed into the deformation standard of corrugated board.It is convenient to make stator-slot wedge models in different states.(2)The test platform of tightness of stator-slot wedges is established and the tightness test is carried out.According to the classification standard of corrugated board shape variables,some test models are made,and the test platform of tightness of stator-slot wedges is built for knocking test.Through the endpoint detection method,the percussion signal is intercepted from 3s to 0.5s,which greatly reduces the amount of calculation for subsequent signal processing.In the practical application of nondestructive detection of stator-slot wedges,the generator repair site is noisy.A noise reduction process is needed to reduce the interference of environmental noise.In this paper,the second generation wavelet transform is used to reduce the noise of the percussion signal,and the noise is reduced.The percussion signals collected in the laboratory simulation test do not need to be processed for noise reduction.(3)The features of the percussion signal are extracted and screened.The features of the pre-processed percussion signal are extracted from time domain,frequency spectrum,and power spectrum.A total of 22 feature parameters were extracted.First of all,according to the different distinctions of the three kinds of tightness by each feature parameter,the 22 feature parameters are divided into 3 groups.Then the F-ratio method was used to screen the sensitivity of the feature parameters in each group and the Pearson correlation coefficient method was used to screen the redundancy between each feature pair in each group.Finally,5 feature parameters are retained.(4)Pattern recognition is performed.The input feature matrix is composed of the above five feature parameters.The test set is classified using the SVM optimized by the cross-validation method to optimize the parameters.The recognition accuracy rate reaches 93.333%.Compared with the recognition accuracy of KNN algorithms and BP neural network,the optimized SVM model has a better effect.This paper innovatively presents the classification standard for the tightness of the stator-slot wedge of corrugated board generators.And non-destructive detection technology based on acoustics has been developed.These have reference value for the research on the tightness detection of the stator-slot wedge of the generator. | | Keywords/Search Tags: | Large hydroelectric generator, Corrugated board, Stator-slot wedge, Tightness, Classification standard, Non-destructive detection, Feature extraction and screening, Support Vector Machine | PDF Full Text Request | Related items |
| |
|