摘要
城市轨道交通牵引能耗占系统整体能耗的40%~60%.车载能耗计量装置记录了大量时间序列形式的牵引能耗数据,其能耗模式和异常分析对城市轨道交通节能运营具有重要意义.本文提出一种针对时间序列数据的牵引能耗异常分析方法.首先基于符号化近似聚合方法降低原时间序列维度得到牵引能耗子模式,然后利用K-means聚类实现牵引能耗模式判别,最后基于获得的典型牵引能耗模式及时间序列相似性度量方法,分析牵引能耗时间序列是否存在异常.将该方法应用于北京地铁某线路列车运营过程中记录的数据,得到3种典型能耗模式并结合车底运用计划进行分析,所提方法可用于时间序列牵引能耗数据的异常判断,为及时发现并确定牵引能耗异常程度提供数据支撑.
In the urban rail transit system, the traction energy consumption accounts for 40%-60% of the total system energy consumption. A large amount of traction energy consumption data is recorded by the on-board energy consumption recorder in the form of time series, whose energy consumption pattern and outlier analysis are of great significance to the energy-saving operation of urban rail transit. This paper proposes a method for outlier analysis of traction energy consumption based on time series data. First, the traction energy consumption sub-patterns are obtained by the symbolic aggregate approximation after reducing the dimension of the original time series. Then, the traction energy consumption patterns are recognized by K-means clustering. Finally, whether the time series of traction energy consumption is abnormal is analyzed based on the typical traction energy consumption pattern and the time series similarity measure. The proposed method is applied to the data recorded on a Beijing Subway during its operation, obtaining three typical energy consumption patterns. And these patterns are analyzed according to the rolling stock scheduling plan. The proposed method can be used to judge the anomalies of traction energy consumption data in time series, providing data support for timely finding and analyzing the anomalies of traction energy consumption.
作者
李熙
张立成
LI Xi;ZHANG Licheng(Beijing Mass Transit Railway Operation Corporation LTD.,Beijing 100044,China;School of Electronic and Information Engineering,Beijing Jiaotong University,Beijing 100044,China)
出处
《北京交通大学学报》
CAS
CSCD
北大核心
2021年第5期30-36,共7页
JOURNAL OF BEIJING JIAOTONG UNIVERSITY
基金
广西科技计划项目(桂科AD20297125)
英国皇家工程院牛顿基金项目(TSPC1025)
北京市交通行业科技项目(201820-JNBJ3)。
关键词
城市轨道交通
牵引能耗
异常分析
时间序列数据
符号化近似聚合
相似性
urban rail transit
traction energy consumption
outlier analysis
time series data
symbolic aggregate approximation
similarity
作者简介
第一作者:李熙(1981—),男,吉林舒兰人,正高级工程师,博士.研究方向城市轨道交通车辆设计、运用与维护.email:bjtulx@163.com.
