| Most of the electricity used in our daily life and production is supplied by distribution lines.However,because distribution lines receive more users and have more complex paths,they are prone to various faults,especially grounding faults.For a long time,although many methods for detecting grounding faults of power lines have been proposed,the detection results are not ideal because of the complex operation environment of distribution lines and many other reasons.Among many fault location methods,traveling wave location method has many advantages,such as high positioning accuracy,not affected by transition resistance and system operation mode,etc.It has been widely used.But most traveling wave location methods only consider the case that only overhead lines or cables are used as transmission lines.However,there are many cases where the overhead line and the cable are mixed in the actual distribution network,and the traveling wave has different wave speed,attenuation degree and dispersion degree on the overhead line and the cable,so there are still many problems.Aiming at the problem that the velocity of wave is difficult to determine,a fault location method based on single-ended traveling wave signal and neural network is proposed.It is not necessary to determine the velocity of modulus wave.By detecting the initial wavefront of single-ended traveling wave modulus and calculating the time difference of modulus,it can verify whether there is a reflection wavefront of corresponding traveling wave in the simulated fault location to realize the fault path determination and location.In view of the complex refraction phenomena of hybrid transmission lines,the magnitude of amplitude attenuation,wave velocity and frequency dispersion of fault traveling waves are not identical,and the initial wave head reflected by fault points is not easy to accurately calibrate the problem.Wavelet transform is used to detect the singularity of traveling waves,and the calibration effect of different wavelet bases for fault traveling waves in specific lines is compared to determine the optimal wavelet basis for solving the problem of frequency dispersion.To solve the dispersion problem,the wavelet transform is used to decompose the traveling wave into different frequency bands,and the specific frequency band is reconstructed to obtain A traveling wave with a small dispersion is used to identify the initial wave head.In order to reduce the influence of noise on traveling wave analysis,a variable threshold denoising method based on the threshold of wavelet coefficients and the number of layers is proposed to eliminate the low-level noise as much as possiblewhile retaining the useful high-level signal,so as to avoid the influence of noise on the subsequent signal analysis.The distribution network model is built on the PSCAD/EMTDC simulation platform and fault traveling wave is collected.The performance of the denoising method in this paper is tested by means of artificial denoising.The different wavelet bases are used to calibrate the wave heads,and the singularity characteristics of the wavelet transform coefficients are compared,and the appropriate wavelet bases are selected.Then the traveling waves are transformed and reconstructed with different layers of wavelet packet,and the optimal number of transform layers is determined according to the fitting effect of the neural network.The fault location results show that this method can achieve fault location in hybrid distribution network with certain precision. |
PDF Full Text Request