Study On Noise Reduction Of Ultrasonic Detection Signal Of Transformer Winding Deformation Based On Wavelet Theory

Winding deformation is the main cause of transformer failure.In order to avoid the occurrence of transformer accidents,the establishment of on-line real-time detection system of transformer winding deformation is of great significance for the timely understanding of the internal winding status of the transformer and the timely discovery of the transformer with potential accidents.When the transformer winding deforms,it will be bulged,displaced or twisted.Therefore,there are corresponding detection technologies of transformer winding,such as short-circuit impedance method,frequency response method,low-voltage pulse method and other off-line detection methods.Although there are mature detection methods and equipment,but in practical engineering applications still need high technical experience,it is difficult to accurately determine the severity of winding deformation.At present,ultrasonic testing technology is the most widely used nondestructive testing method to ensure the safe operation of equipment.Therefore,the use of ultrasonic detection technology to establish the transformer winding deformation online real-time detection system is conducive to the operation monitoring and efficient maintenance of the transformer,to meet the development requirements of the implementation of condition based maintenance in the power industry.After the transformer winding deformation on-line real-time detection system is established,due to the inevitable existence of various noises in the actual detection process,the system has hardware to amplify and filter the ultrasonic detection signal of transformer winding deformation,but the noise reduction effect can not meet the requirements of engineering application,There is still a lot of noise in the ultrasonic detection signal of transformer winding deformation transmitted to the digital signal processor,which makes the useful information of transformer winding deformation in the ultrasonic detection signal polluted.Therefore,it is of great significance to reduce the noise of the ultrasonic detection signal of transformer winding deformation from the algorithm.Based on the principle of transformer winding deformation ultrasonic detection,wavelet theory and the characteristics of transformer winding deformation ultrasonic detection signal,this paper studies the application of wavelet theory wavelet threshold noise reduction method in transformer winding deformation ultrasonic detection signal noise reduction.In view of the shortcomings in the selection of wavelet basis function,threshold function and threshold value in the traditional wavelet threshold denoising method,the selection and improvement method is proposed.In terms of the selection of wavelet basis function,according to the characteristics of wavelet basis function,through analysis and comparison,we finally choose the Sym8 wavelet which is similar to the ultrasonic detection signal waveform of transformerwinding deformation.In the aspect of threshold function selection,aiming at the shortcomings of traditional threshold function,an improved threshold function is proposed,and simulation noise reduction is carried out.Simulation results show that the improved wavelet threshold noise reduction method can achieve better noise reduction effect.In the aspect of threshold selection,aiming at the shortcomings of common threshold selection criteria,an improved threshold selection is proposed.Considering that in the practical application of the transformer winding deformation online real-time detection system,the noise in the ultrasonic detection signal of transformer winding deformation collected by the digital signal processor is almost Gauss white noise,so the one threshold readjustment method is adopted.Under this threshold readjustment method,the ultrasonic detection signal of transformer winding deformation with noise is processed and analyzed by simulation The results show that the improved threshold is better.It provides a theoretical basis for the noise reduction of ultrasonic detection signal of transformer winding deformation in the on-line real-time detection system of transformer winding deformation.