Traveling Wave Fault Location In Fully Parallel At Traction Network Based On Section Decomposition

The full parallel AT traction network undertakes the power supply task for electric locomotives.Due to the complex line topology and environment along the line,it is often prone to short-circuit fault.Therefore,fast and accurate fault location is particularly important.The traveling wave fault location technology is becoming more and more mature in transmission lines,so the traveling wave method will be used to study the fault location of all-parallel AT traction network.Due to the parallel connection of all-parallel AT traction network at AT through the transverse line,when the short-circuit fault point occurs at different AT section intervals,the transient traveling wave characteristics monitored by the power supply arm port are quite different,and the wave head is difficult to distinguish.To this end,a traveling wave ranging method based on section decomposition is proposed in this paper.This method preliminarily determines the fault point of traction network by training the deep learning section model,and then selects the data of the measuring points at the corresponding section end,and uses the Atype traveling wave ranging method to calculate the fault distance.Firstly,in view of the limited acquisition of short-circuit fault data of traction network on site,the electromagnetic transient simulation software ATP-EMTP is used to establish the shortcircuit fault model of traction network under different conditions.Secondly,aiming at the problem of selecting the traveling wave ranging method suitable for traction network and the wave velocity and the second wave head identification under the selected method,combined with the theoretical and simulation analysis of the propagation characteristics and laws of traveling waves,the traveling wave ranging method suitable for traction network is proposed.Then,the current phase-mode transformation matrix is derived to analyze the frequency characteristics of traveling wave modulus,calculate the wave velocity of each modulus,and select the appropriate modulus for ranging.Thirdly,aiming at the problem that is difficult to identify the second reflected wave head in the A-type traveling wave location principle under the selected traveling wave location,a section location method based on deep learning is proposed to preliminarily determine the fault section,so as to complete the property identification of the second reflected wave head.Finally,a traveling wave ranging method based on segment decomposition is proposed and used for fault location of fully parallel AT traction network.The wavefront is calibrated by using the modulus maximum to determine the wavefront moment.The simulation results show that the proposed method can achieve accurate positioning.