随机森林在轨道交通车辆门系统亚健康状态诊断中的应用
严俊1支有冉2许志兴2史翔2
Application of Random Forest in the Sub\|health Diagnose of Railway Vehicle Door System
YAN JunZHI YouranXU ZhixingSHI Xiang
-
作者信息:1.1. 上海地铁维护保障有限公司，200070,上海;
2.2. 南京康尼机电股份有限公司，210013,南京
-
Affiliation:Shanghai Metro Maintenance and Guarantee Co., Ltd., 200070, Shanghai, China
-
关键词:
-
Key words:
-
DOI:10.16037/j.1007-869x.2019.12.027
-
中图分类号/CLCN:U279.3+23
-
栏目/Col:研究报告
摘要:
提出了一种基于随机森林(RF)的车门亚健康状态诊断方法。首先，从车门驱动电机的电流、转速和转矩输出信号中提取时频域特征指标作为表征车门运行状态的特征向量；然后，RF通过对分类器的集成来提高车门亚健康状态诊断的精度，并且从特征指标对分类结果影响的角度评估特征的重要性。利用台架调门试验模拟车门的典型亚健康工况，采集数据并分析验证。试验结果表明，与传统的单分类器（决策树、支持向量机、贝叶斯、KNN近邻）的诊断结果相比，RF方法在车门的亚健康状态诊断中具有更高的诊断精度，并可对特征的重要性进行评估。
Abstracts:
The sub\|health status diagnose of railway vehicle door system (RVDS) based on Random Forest (RF) is proposed. Firstly, features in time domain and frequency domain are extracted from the output signal of the door driving motor, including current, speed and torque as the feature vectors of RVDS operation; secondly, by integrating the classifiers, RF will improve the accuracy of sub\|health diagnosis and evaluate the importance of features, from the perspective of feature vectors impact on the final classification. And finally, the bench adjustment experiment is used to simulate the typical sub\|health conditions of RVDS, data are collected for analysis and verification. The results show that compared to the traditional single classifier (decision tree, support vector machine, Bayes, KNN algorithm), RF method has higher diagnostic accuracy in RVDS sub\|health diagnose, and can assess the importance of features.
过刊浏览
资料下载
友情链接