摘要
以1970-2008年我国人均生活电力消费量作为原始数据序列,首先应用SCGM(1,1)_c模型模拟原始序列的总体趋势;然后将所得到的相对误差作为随机波动过程,将原始序列的归一化自相关系数作为权重,应用Markov链原理预测2009年的状态,进一步预测2009年电力消费量,并与实际数据比较,检验预测精度;同样地,应用等维新陈代谢思想,对2010-2012年电力消费量进行了预测,并检验预测精度,达到了滚动建模和动态预测的目的.结果显示,等维新陈代谢-加权Markov-SCGM(1,1)_c模型的平均模拟精度为98.3%,平均预测精度为96.0%.最后对2013-2017年我国人均生活电力消费量进行了预测.
We firstly apply SCGM(1,1)c model to simulate the general tendency of the original sequence basing China's per capita consumption of electricity from 1970 to 2008 as the original statistical sequence. We secondly use the comparative error as undulation stochastic process, regard the normalizing autocorrelation coefficient of original sequence as weight, apply the principle of Markov Chain to predict the con- sumption of 2009, and compare the result with the actual data to ensure the exactness. Similarly, in terms of dimension equality metabolism, the prediction of electricity consumption from 2010-2012 is conducted and checked to achieve the goals of rolling modeling and dynamic prediction. The result demonstrates that the average simulating exactness of dimension equality metabolism-weighted MarkowSCGM(1,1)c has reached up to 98.3%, the average prediction exactness 96.0%. Finally, prediction of China's per capita electricity consumption from 2013 to 2017 is made.
出处
《系统科学与数学》
CSCD
北大核心
2014年第5期521-533,共13页
Journal of Systems Science and Mathematical Sciences
关键词
人均生活电力消费量
单因子系统云灰色模型
加权马尔柯夫链
最优分割法
预测.
Average power consumption of living, SCGM(1,1)c model, weightedMarkov Chain, optimum partitioning clustering method, prediction.