一种基于贝叶斯岭回归的制动机气动阀故障预警方法

打开文本图片集

中图分类号：U270.35 文献标志码：B doi：10.20213/j.cnki.tdcl.2025.01.014

Abstract： As an important component in the train electro-pneumatic braking system，the fault alert of pneumatic valves is of great significance to ensure the safetyof train operation.Considering thefaults of rubber ring wear，spring aging and friction increase in valve disk，this paper proposes a fault alert method based on Bayesian Ridge Regression. A joint simulation model of AMESim and MATLAB is established to generate the whole life cycle data of pneumatic valves，and the principal component analysis method is used to extract health indicatorsfrom the data set，and the fault prediction model based on Bayesian Ridge Regression is trained to accurately estimate the occurrence time of pneumatic valve failure. A hardware-in-the-loop simulation verification platform of train tube/balance control based on DK-2 brake is built，and the feasibility and effectiveness of the proposed method are verified by progresive fault simulation and measured data analysis of the experimental platform.

Key words： braking system；pneumatic valve；fault alert；principal component analysis；Bayesian Ridge Regression

1概述

气动阀作为列车制动系统“电路-空气-机械"控制回路的关键部件，对系统的安全性和可靠性起到至关重要的作用[]。（剩余8544字）