出于提高天然气政策效率、优化天然气消费预测以及改善天然气管网规划等目的，运用基于模糊等价关系闭包的聚类方法对大陆30个省份(不含西藏) 2013年天然气消费结构进行了聚类研究。结果为，这些省份：1) 在0.9的精度水平上，共有22类，其中河北与黑龙江组成1类，山西与湖北组成1类，江西、广东、贵州、云南与陕西组成1类，其余省份各独为1类；2) 在0.8的精度水平上，共有12类，其中北京、天津、内蒙古、上海、福建、河南、海南、重庆、青海及新疆各独为1类，江苏与山东组成一类，其余省份组成1类；3) 在0.7的精度水平上，共有4类，其中重庆独为1类，北京与上海组成1类，福建、河南组成1类；4) 在0.6的精度水平上，共有2类，其中重庆独为1类，其余所有省份组成1类；5) 在0.5的精度水平上，仅有1类。据研究结果，得三个主要结论：1) 地域特征并非中国天然气消费结构差异的唯一因素；2) 重庆、北京、上海、福建以及河南在天然气消费结构方面难以与其它省份聚为1类；湖北与山西以及江西、广东、贵州、云南与陕西在这方面很大程度上又各同属一类；3) 重庆表现出了最明显的独特性。这些现象的深层原因值得进一步研究。
In order to enhance the efficiency of natural gas policy, optimize the forecast of natural gas con-sumption and improve the plan of natural gas pipeline construction, an fuzzy clustering based on the fuzzy equivalence relation’s transitive closure is applied to the natural gas consumption structures of 30 provinces (excluding Tibet) of Mainland China in 2013. The results show that, among those provinces: 1) at the accuracy level of 0.9, there are 22 clusters, with Hebei and Hei-longjiang being one cluster, Shanxi and Hubei being another, Jiangxi, Guangdong, Guizhou and Shannxi being a third and each of the rest along being a single cluster; 2) at the accuracy level of 0.8, there are 12 clusters, with Beijing, Tianjin, Inner Mongolia, Shanghai, Fujian, Henan, Hainan, Chongqing, Qinghai and Xinjiang each being a single cluster, Jiangsu and Shandong being one, and all others being another; 3) at the accuracy level of 0.7, there are 4 clusters, with Chongqing along being a single, Beijing and Shanghai being one, Fujian and Henan being another, and all others be-ing the fourth; 4) at the accuracy level of 0.6, there are 2 clusters, with Chongqing along being a single, and all others being the other; and 5) at the accuracy level of 0.5, there is only 1 cluster. Three points can be concluded from the results. First, no evidence of regional characteristic has been found associated with the structures of natural gas consumption in those provinces. Second, it turns out to be rather difficult for Chongqing, Beijing, Shanghai, Fujian as well as Henan to be classified in one cluster, while Hubei & Shanxi and Jiangxi & Guangdong & Guizhou & Yunnan & Shannxi belong to one cluster rather significantly. And third, Chongqing appears the most unique. The in depth logics of those phenomena are worth further investigation.
Yu, S., Wei, Y.-M., Fan, J., Zhang, X. and Wang, K. (2012) Exploring the Regional Characteristics of Inter-Provincial CO2 Emissions in China: An Improved Fuzzy Clustering Analysis Based on Particle Swarm Optimization. Applied Energy, 92, 552-562. http://dx.doi.org/10.1016/j.apenergy.2011.11.068