Weak Signal Detection And Fault Degree Judgment Under Online Operaton Of Transformer Winding

Power transformer is the core equipment of power transmission and transformation in power system,and fault state detection of its winding is very important for power supply reliability.On-line detection can find the winding fault status in real time,but the fault signal will be submerged under the background of strong noise.Aiming at the problem that it is difficult to extract the fault signal from online detection,this paper uses the orthogonal vector detection method based on the correlation algorithm to study,and uses the correlation coefficient method to judge the transformer winding deformation.First,it introduces the principle of weak signal detection related algorithms and the principle of improved orthogonal vector algorithm.The MATLAB software is used to simulate and compare the anti-interference ability of the two algorithms on the fault signal under three background noises.The detection ranges of the cross-correlation algorithm and the orthogonal vector algorithm are respectively SNR>-25.5dB and SNR>-26.7dB under Gaussian white noise;SNR>-36.3dB and SNR>-41.4dB under power grid noise;Under mixed noise,SNR>-26.4dB and SNR>-26.7dB respectively.The simulation results show that the orthogonal vector algorithm has a wider detection range than the cross-correlation algorithm.It was verified by field data,and the results showed that the use of quadrature vector algorithm can extract the useful signal of μV from the noise-containing fault signal with voltage level of mV,and the detected amplitudes are 3.912μV,3.836μV,3.945μV,4.012μV,the experimental errors are-6.205%,-3.422%,7.435%,2.193%,which are all less than 10%,which meets the requirements of the power industry standard detection error of less than 20%(1dB),indicating that the orthogonal vector algorithm can be Used in online transformer winding fault signal detection.Secondly,using PSpice simulation to analyze the effect of winding deformation on the equivalent model parameters and the frequency response characteristic curves under different deformations.The research results show that:when the winding deforms,the peak and valley of the frequency response characteristic curve(resonance point)move to the high frequency side;when the winding deforms,the peak and valley of the frequency response characteristic curve move to the low frequency side.The correlation coefficient method is used to study the amplitude-frequency characteristic curve of the actual winding,and the calculation result is verified by experiments.The correlation coefficients obtained for the low frequency band,mid frequency band and high frequency band are 0.8123,0.5974,and 0.5330 respectively.According to the relevant standards of the power industry,the judgment result is consistent with the actual transformer winding deformation,indicating that the calculation result can be applied to actual detection.Finally,a software system for transformer winding deformation monitoring based on Lab VIEW software is developed.The system mainly includes:signal extraction module using orthogonal vector algorithm;using correlation coefficient method to calculate characteristic value of winding deformation frequency response characteristic curve and characteristic value fault judgment module;result display module.And the sub-modules were verified experimentally by collecting data on the spot.The verification results show that the operating results of each module are consistent with the actual results,indicating that the method selected in the design is reasonable,the software programming is correct,and it can be used in actual projects.