基于EEMD弓网检测数据相关性的接触压力数据分析  

Pantograph-catenary Contact Force Data Analysis Based on Data Correlation Decomposed by EEMD

作　　者：韩志伟[1] 刘志刚[1] 张晓晓[1] 刘煜铖[1] 

机构地区：[1]西南交通大学电气工程学院,四川成都610031

出　　处：《铁道学报》2013年第9期25-30,共6页Journal of the China Railway Society

基　　金：国家自然科学基金(U1134205;51007074);铁道部科技研究开发计划(2011J016-B);中央高校基本科研业务费专项资金资助项目(SWJTU11CX141)

摘　　要：为提高弓网检测数据分析的准确性,提出一种利用多元弓网检测数据间相关性进行接触压力数据预处理的方法。为克服实测数据变化复杂、难以直接度量数据相关性的问题,利用EEMD对多元检测数据进行IMF分解;通过定义奇异性突变比对IMF分量进行预处理;最终采用局部皮尔逊相关系数及局部时延皮尔逊相关系数评估检测数据IMF分量间的数据相关性。对多元检测数据与接触压力数据的相关性系数研究表明:接触压力数据中超过3倍标准差的数据包含了检测错误点、硬点和线路参数超标点3类,通过剔除记录错误点,保留硬点和弓网参数造成的超标点,可为更加准确评估弓网工况及分析接触线状态缺陷提供帮助。A pantograph-catenary contact force data pretreatment method by analyzing detection data correlation was proposed to improve the data analysis accuracy.The data were decomposed by ensemble empirical mode decomposition （EEMD） first to simplify the complex data comparison.Then,the pretreatment called the singularity mutation ratio（SMR） was applied to the intrinsic mode functions （IMF）.The IMF component correlations were assessed by the local Pearson correlation coefficient and delayed local Pearson correlation coefficient at last.Study on the correlation coefficient between the multi-detection data and contact force data shows that the over-standard contact force data exceeding three times the standard deviation include the error data,hard spots data and over-standard data caused by catenary parameters.The assessment of pantograph-catenary performance and contact wire defect status are more accurate after deleting the error data and reserving the data of hard spots and over-standard spots.

关 键 词：接触网 相关性 数据预处理 聚类经验模态分解 

分 类 号：U226.8[交通运输工程—道路与铁道工程]