Investigation Of Nanosecond Impulse Frequency Response Analysis For Power Transformer Winding Deformation And Displacement Detection

Power transformer is a major apparatus in a power system.According to statistics,winding faults caused by short circuit currents due to windings' lack of mechanical strength is the most common cause of power transformer failures.The first symptom of serious winding faults is usually winding deformation and displacement which is usually harmless at first but becomes greater and greater after the occurrence of another and yet another short circuit current,and finally cause a failure of a transformer.Therefore,to detect winding deformation and displacement in an early stage and to protect it from becoming more serious can reduce the power transformer failures and thus,improve the operation stability of a power system.The detection of winding deformation and displacement is still offline,however,in order to detect it in an early stage,it is necessary to develop online methods.Nanosecond Impulse Frequency Response Analysis(Nanosecond IFRA)has the potential to realize online monitoring of transformer winding,because it is fast,the frequency band of it is easy to extend,it has little influence on the operation of the power system and so on.But this method is not mature,the impulse and the sampling parameters of which have not been decided yet and the reliability,sensitivity and reproducibility has not been verified.For the application of Nanosecond IFRA the main contributions of this research is listed as followed:(1)The relationship between the frequency resolution and signal to noise ratio of the Nanosecond IFRA frequency response curve obtained through the Fast Fourier Transform(FFT)is stated.To reduce noises on the Nanosecond IFRA frequency response curve without reducing the frequency resolution,the reference for parameters of the impulse and sampling is given based on the experimental results.To further reduce the noises,considering the distribution of resonance points and noises,a novel denoising method based on threshold wavelet is designed.(2)To verify the reliability and sensitivity of Nanosecond IFRA,this method and a mature method,Sweep Frequency Response Analysis(SFRA)are used to detect the same three levels of winding displacement on a dry-type distribution transformer.To verify the reproducibility of this method,the same experiment is conducted 30 days later.The results are analyzed by statistical parameters.