Train motion state estimation method based on few-sample-element lifting learning

Abstract

The invention relates to a train motion state estimation method based on few -sample -element lifting learning, and belongs to the technical field of computer and information science. The method mainly solves the problems that an existing train motion state modeling method has the high-cost modeling problem whether a physical model or a machine learning model exists, online continuous self-adaption of the model is difficult to achieve for a specific train so as to achieve accurate simulation, systematic simulation errors exist, and application requirements of real-time simulation and the like of the automatic train driving system are difficult to meet. According to the method, firstly, a model is established by adopting a meta-gradient lifting learning algorithm on the basis of metadata, and then task model learning is completed by adopting a task gradient lifting learning algorithm on the basis of a small amount of data for a new task, so that rapid and low-cost accurate simulation of a new train is realized. The result shows that the train motion state can be accurately estimated, the training cost of the model is reduced, and the accuracy of train motion state estimation is improved.

Description

technical field [0001] The invention relates to a method for estimating a train motion state using a few-sample element boosting learning, and belongs to the technical field of computer and information science. Background technique [0002] Train motion state estimation can be oriented to a real train in motion or a simulated train in a simulation environment. Based on a specific control command sequence, the estimation or prediction of its speed, position and other indicators is completed, which is the basis for the establishment and operation of a train automatic driving system. Motion state estimation is generally a continuous time series estimation, such as simulating a train with a speed of zero at the starting point, passing in control commands cycle by cycle, and simulating the whole process of acceleration, cruising, deceleration, etc., until the train reaches near the end point and the velocity is zero. In a single cycle, the input of the estimation model may inclu...

AI Technical Summary

Problems solved by technology

[0015] The purpose of the present invention is to solve the problem of high-cost modeling in the existing modeling methods of train motion state, whether it is a physical model or a machine learning model. simulation error, and it is difficult to meet the application requirements of real-time simulation of train automatic driving system, a method of accurate modeling of train motion state element promotion learning is proposed

Images