Data-based Analysis on Traction Energy for Urban Rail Transit Using Clustering and Classification Methods

• Published in: 2019 IEEE 8th Data Driven Control and Learning Systems Conference (DDCLS) pp. 191 - 196
• Main Authors: Zhao, Jiao, Tang, Tao, Xun, Jing, Jiang, Shan, Gao, Ming, Ren, Baifeng
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2019
• Subjects: Automobiles; Buildings; Clustering algorithms; Energy consumption; Energy Consumption Patterns; Outlier Detection; Public transportation; Rails; Traction Energy Consumption; Traction motors; Urban Rail Transit
• DOI: 10.1109/DDCLS.2019.8909030

The energy consumption of urban rail transit is increasing gradually in recent years. Accurate data analysis of traction energy is in urgent demand for improving energy-efficiency, thus it becomes an essential research subject for the metro system. According to the characteristics of the subway, this paper analyzes the traction energy consumption data based on the clustering and classification methods. The energy data collected by the energy consumption metering device is a line from Beijing Subway. The energy data feature vectors are obtained after simply pre-processing the raw data. Cluster algorithms are applied to divide the feature vectors into several sets with the same characteristics. The energy patterns is generated by decision tree algorithm. Finally, the outliers of traction energy under the same energy pattern are selected based on local outlier factor (LOF). The analysis results show that the traction energy data can be divided into four patterns and the outliers are detected under finer clusters than before, which can help the metro companies manage the energy data delicately.