高斯过程回归在轴承健康状态预测中的应用
Application of Gaussian Process Regression in Prediction of Bearing Health
马浩 1石永进 1李伟1
作者信息
- 1. 西安中车永电捷力风能有限公司,陕西西安 710086;轨道交通牵引电机山西省重点实验室,山西永济 044502
- 折叠
摘要
轴承作为列车牵引电机的重要零部件,准确评估其后续健康状态对列车的安全运行至关重要.基于轴承台架试验及实车数据,将主成分分析(PCA)和高斯过程回归(GPR)应用于轴承的健康状态评估及预测中.通过对试验数据进行特征提取,获得能够表征轴承衰退规律的特征数据,应用PCA将提取的数据进行降维并建立轴承状态评估数据,应用GPR对评估数据进行学习和预测.通过预测值与真实值的对比验证,GPR可以实现轴承的健康状态预测,并在低采样率下保持了较高的准确率.
Abstract
Bearing is an important part of train traction motor and accurate evaluation of its subsequent health status is very important for train safe operation.Based on the bearing bench test and real vehicle data,principal component analysis(PCA)and Gaussian process regression(GPR)are applied to assese and predict bearing health status.Features are extracted from the experimental data,obtaining the characteristic data that can characterize the law of bearing degradation.PCA is used to reduce the dimension of extracted data and establish the bearing state assessment data,and GPR is applied to learn and predict the assessment data.Through the comparison of the predicted value and the true value,GPR can predict the health of bearing and maintain a high accuracy rate at a low sampling rate.
关键词
轴承状态预测/高斯过程回归/主成分分析/牵引电机Key words
bearing condition prediction/Gaussian process regression/principal component analysis/traction motor引用本文复制引用
基金项目
国家重点研发计划项目(2020YFB2007905)
出版年
2024
国家科技期刊平台
NSTL
万方数据