Data-driven relative position detection technology for high-speed maglev train

Abstract

High-speed maglev train speed measurement and positioning system is the basis of train traction and operation control. The existing relative position based on the principle of electromagnetic induction is vulnerable to electromagnetic interference and suspension gap fluctuations, which requires a lot of time to solve electromagnetic compatibility and other issues. In this paper, a data-driven relative position detection scheme based on laser displacement sensor combination is proposed for the high-speed maglev train speed measurement and positioning system. It uses a combination of laser displacement sensors to collect the vertical displacement signal and establishes the relative position detection model of one coggle cycle by designing a neural network to sense the periodic change of vertical displacement signal, which effectively overcomes the influence of suspension gap fluctuation. In order to improve the modeling accuracy, an adaptive fault-tolerant filtering algorithm is proposed, which can effectively overcome the effects of inconsistent width of the long stator surface, inconsistent depth of the stator cable depression, and foreign bodies on the stator surface. At the same time, the proposed scheme can determine the rail gap in real time, through the two detection signal switching mode can eliminate the influence of rail gap. The analysis and experimental results show that the proposed solution can effectively adapt to various conditions of vehicle operation, overcome the influence of electromagnetic interference and suspension fluctuations, and realize the high-precision positioning requirements of high-speed Maglev trains.