用可拓聚类模型预测天然气的需求  

Prediction of natural gas demand based on extension classified model

作　　者：黄大富[1] 任竞争[1] 江寒梅[1] 谭世语[1] 董立春[1] 周志明[1] 何思然[1] 

机构地区：[1]重庆大学化学化工学院,重庆400030

出　　处：《计算机与应用化学》2011年第5期617-619,共3页Computers and Applied Chemistry

基　　金：中央高校基本科研重庆大学研究(CDJXS10221140);重庆大学211工程三期创新人才培养项目(S-09103)

摘　　要：天然气作为1种广泛使用的能源,其需求的预测对工业生产具有十分重要的指导意义。目前报道的天然气需求量预测模型或所需样本太多,或缺乏严谨的建模机理,因此,建立1种具备严格理论基础且只需少量样本就可准确预测的模型非常重要。建立基于可拓理论和聚类分析的天然气需求预测模型,将工业用气量、GDP增长率、燃料价格指数作为影响天然气需求量的主要因素,在对历史数据分析的基础上,构造出物元模型的经典域和节域,再根据关联函数,确定待测物体元对各个经典域的关联度,判断待测样本变化率所属类别,进而预测待测年份天然气需求量的范围。根据1990年至2001年的天然气需求量及其影响因素的相关数据,对2002年天然气预测量进行预测,预测结果为2002年天然气需求量相对2001年的变化率为1.06～1.08之间,与真实值1.0771非常接近。可见可拓聚类预测模型可以作为1种新的可准确预测天然气需求量的方法。Natural gas is a widely-used energy resource,the accurate prediction for natural gas demand is very important to guide the industrial production. While the reported models either require large amount of samples or lack strict modeling mechanism,so it is very useful to establish a model that is built upon a strict theoretical basis and requires a few amount of samples for accurate prediction.In this paper,an extension classification model was established for the prediction of natural gas demand,which takes account of industrial consumption of natural gas,GDP growth rate,and price index of fuels as the main affecting factors of natural gas requirement By analyzing the historical data,the classical and controlled domains of the material elements were constructed,then the correlation of every material element to the classical domains was established according to the correlation functions,which was used to decide the category for the variance rate of the target sample and further predict the range of the natural gas demand for the target years.According to the data of natural gas demand and the relative affecting factors from 1990 to 2001,the natural gas demand in 2002 was forecasted using the established model,the growth rate was predicted to be between 1.06 and 1.08,which is very close to the actual value,1.0771.Therefore,it demonstrated that the extension classification model can provide a new method that can accurately forecast the natural gas demand.

关 键 词：天然气需求量 可拓聚类预测 物元 关联函数 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]