Train emergency parking method based on small sample data enhanced ensemble learning

Abstract

The invention provides a train emergency parking method based on small sample data enhanced ensemble learning, which is characterized in that a small sample data set is utilized to perform data enhancement through a GAN (Generative Adversarial Network), a braking distance prediction ensemble learning model is established through stacking ensemble learning, and an evolutionary algorithm is utilized to perform adjustment and optimization. When the train encounters an obstacle, the maximum service braking distance is predicted by utilizing the braking distance prediction integrated learning model, so that whether the maximum service braking torque is adopted for braking or the maximum emergency braking torque is adopted for braking is accurately judged, and the personal safety and the riding comfort of passengers are ensured to the maximum extent on the premise of ensuring the safe running of the train.

Description

technical field [0001] The invention relates to the technical field of transportation, in particular to a train emergency parking method based on small sample data enhanced integrated learning. Background technique [0002] For the rail train running lines in some mountainous cities, the slope problem is more prominent, and it is easy to cause geological disasters such as dangerous rocks and landslides. , forming accumulation layer landslides of varying scales, posing a serious threat to the safety of urban rail lines. In the prior art, obstacle detection elements such as cameras or lidars in the on-board driving control system can detect obstacles that appear on the train line, and then calculate the distance between the current position of the train and the obstacle. There are usually two braking methods to choose from: the maximum common braking torque braking with a longer braking distance and the emergency braking torque braking with a short braking distance. The disc...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problem of the background technology, the present invention provides a train emergency parking method based on small sample data enhanced integrated learning to solve the difficulty in judging or correctly selecting the braking and parking mode when the train encounters obstacles during the running of the train in the prior art, so as to The problem of ensuring the safety of trains while ensuring the safety of passengers

Images