Empirical Study on Abnormal CNG Refilling Behaviors Based on Single Attribute

2018 2nd International Workshop on Renewable Energy and Development

能源学;经济学

Li, Yang^1,3 ; Zhao, Yulian^1 ; Jin, Dengchao^1 ; Sun, Jinhui^2 ; Shao, Jian^3

College of Engineering and Technology, Tianjin Agricultural University, Tianjin

300384, China^1

Tianjin Key Lab of Aqua-Ecology and Aquaculture, College of Fisheries, Tianjin Agricultural University, Tianjin

300384, China^2

Tianjin Huabei Gas and Heating Engineering Design Co. LTD, Tianjin

300384, China^3

关键词: Abnormal behavior;    CNG vehicles;    Empirical studies;    Final pressure;    Initial pressure;    Kernel Density Estimation;    Probability densities;    Unsupervised learning method;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/153/3/032053/pdf
DOI  :  10.1088/1755-1315/153/3/032053

来源: IOP

【 摘 要 】

This study used mainly boxplot method to detect abnormal CNG refueling behaviors. The ownership of CNG vehicles in China has ranked the first in the world for many years. However, due to difficulties such as approval of land, CNG filling stations are difficult to meet the increasing demand for refueling. Therefore, the boxplot method was used to study more than 0.25 million CNG refilling records from a secondary CNG filling station in Tianjin to improve the refilling efficiency in this article. The study found that there were about 2.5% abnormal refilling behaviors based on volume, initial pressure and final pressure separately. To verify those results, another unsupervised learning method of kernel density estimation was used to estimate the probability density of each attribute. Finally, it was found that repeated refilling actions were the main reasons of abnormal behaviors detected based on volume attribute. The results suggested that the way to reduce abnormal CNG refilling behaviors was to avoid repeated refilling actions.