A train condition on-line prediction method based on long-short-time memory network

Abstract

The invention discloses a train condition on-line prediction method based on a long-short-time memory network. Firstly, the train condition monitoring data is pretreated and normalized to complete theselection and division of the condition monitoring data. Then the LSTM is designed as a network chain structure, and a multi-layer interactive structure is established. The structure of LSTM networkgate is designed, and the LSTM unit structure including input gate, forgetting gate and output gate is obtained. The back propagation algorithm based on solving gradient is used to update the networkweight and bias parameters. Finally, the train speed data is selected as the research object and the train speed prediction is carried out. The invention solves the problems that the train state prediction in the prior art is inaccurate, the fault false alarm rate is high, the traditional estimation algorithm operation is complex, the portability is poor, the off-line state prediction can not grasp and control the real-time state of the train, and the prediction precision is not high. The invention solves the problems that the train state prediction is inaccurate, the fault false alarm rate ishigh, the traditional estimation algorithm operation is complex, the portability is poor.

Description

technical field [0001] The invention belongs to the technical field of rail transit operation safety, and in particular relates to an online train state prediction method based on a long-short-term memory network. Background technique [0002] With the rapid development of trains in our country, more and more attention has been paid to the safety of trains. In order to promote the healthy and sustainable development of my country's rail transit, it is very important to carry out real-time online prediction adapted to the train status. [0003] Accurate train state prediction is the basis for ensuring high-speed and safe operation of trains. Through the online real-time prediction of the train status, the real-time performance of each part of the train can be grasped, laying the foundation for the real-time and precise control of the train. Therefore, it is of great significance to carry out basic research on train status prediction based on Long Short Term Memory Network (...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide an online train state prediction method based on long-short-term memory network, which solves the inaccurate train state prediction in the prior art, high fault false alarm rate, complex operation of traditional estimation algorithm, and poor portability. The offline state prediction cannot grasp and control the real-time state of the train, and the prediction accuracy is not high

Images