The autotransformer (AT) neutral current ratio method is widely used for fault location in the AT traction power network. With the development of high-speed electrified railways, a large number of data show that the relation between the AT neutral current ratio and the distance from the beginning of the fault AT section to the fault point (Q–L relation) is mostly nonlinear. Therefore, the linear Q–L relation in the traditional fault location method always leads to large errors. To solve this problem, a large number of load-related current data that can be used to describe the Q–L relation are obtained through the load test of the electric multiple unit (EMU). Thus, an improved fault location method based on the back propagation (BP) neural network is proposed in this paper. On this basis, a comparison between the improved method and the traditional method shows that the maximum absolute error and the average absolute error of the improved method are 0.651 km and 0.334 km lower than those of the traditional method, respectively, which demonstrates that the improved method can effectively eliminate the influence of nonlinear factors and greatly improve the accuracy of fault location for the AT traction power network. Finally, combined with a short-circuit test, the accuracy of the improved method is verified.
Read full abstract