Precise Detection Method Of Traveling Waves Based On Improved Step Size LMS Adaptive Algorithm

In the traveling wave signal generated by the power grid fault,the high-frequency part contains rich fault information,and many high-frequency fault information has the incomparable advantages of power frequency fault information.As a typical protection based on transient component analysis,traveling wave technology It is one of the technologies that has been widely studied in various protection fields.Due to the limitation of the existing measurement technology level,it is impossible to directly obtain the primary traveling wave signal of the grid-side fault,and can only measure the secondary signal after the transformation through the transformer and other devices for traveling wave positioning.However,after the primary traveling wave signal of the power grid is transmitted through transformers,detection modules and other devices,the waveform will be distorted to a certain extent.Therefore,it has important theoretical and practical application value to solve the problem of traveling wave distortion caused by traveling wave sensor,obtains the real primary traveling wave signal,effectively improve the authenticity of traveling wave detection,and promote the practical application of traveling wave technology.The thesis firstly studies the working principle of Rogowski coil and analyzes the transmission characteristics of traveling wave sensor.It is verified by experiments and simulations that after the fault traveling wave signal is transmitted by the transformer,a certain degree of waveform distortion will occur.Aiming at this problem,two solutions are discussed,one is to study the internal structure of the sensor to reduce the degree of distortion,and the other is to use the secondary side traveling wave signal to reversely derive the primary side signal to obtain the real primary side signal.At present,the second method is mostly used to obtain the actual fault transient signal,and this reverse derivation process is a typical inversion problem solving process.For the signal inversion problem,the inverse function can be obtained based on the forward transfer function of the traveling wave sensor model,but this method requires accurate modeling of the sensor.Once the sensor model parameter settings are deviated,it will eventually Affects the solution and inversion accuracy of the inversion signal.In order to solve the problems existing in the inversion based on the transfer function method,the paper proposes an inversion algorithm based on adaptive filtering,discusses the two most common adaptive algorithms,and analyzes and improves the learning rate parameter and performance to improve the convergence.Finally,an improved step-size LMS adaptive inversion algorithm is proposed.The method firstly builds the IEEE 9 node standard test model and sensor model,obtains the primary and secondary traveling wave signals of the transmission line under different fault locations,initial phase angles,transition resistances and different fault types,and then uses the variable-step LMS adaptive algorithm.Train the primary and secondary signals,find the transformation relationship between the primary and secondary signals,and establish a black box inversion model.Considering the existence of noise in the actual secondary traveling wave signal,wavelet threshold denoising is introduced to predict the detected secondary traveling wave,and finally the black box model is used to invert to obtain the primary traveling wave.The simulation and experimental results show that the method does not need to use the sensor transfer function to obtain the inversion model,which can avoid the influence of inaccurate transfer function on the inversion results,effectively improve the authenticity of traveling wave detection,and promote the practical application of traveling wave technology.