摘要
鉴于我国能源消费系统的复杂性及非线性特征,分别采用神经网络和时间序列两种方法建立我国能源消费总量的单项预测模型,并对各模型进行了检验,模型的检验结果表明建立的模型可以作为预测未来能源消费量的有效工具。根据标准差法对各模型的结果进行权重分配,建立我国能源消费的组合预测模型,组合预测模型既克服单一模型的缺陷,又提高了预测精度,之后应用此模型对我国未来六年的能源消费进行预测,2015年我国能源消费总量将会达到41.9亿吨标准煤。
Considering the complexity and nonlinear characteristics of China＇s energy consumption system,neural networks and time series are used to establish individual forecasting models for China＇s energy consumption system,and each of the models was tested. The results showed that the models could be used as effective tools to predict China＇s future energy consumption. According to standard deviation method,suited weight was distributed to the prediction of each individual model,then a combination forecasting model was established. The combination model not only get rid of defects of the former models,but it raised the accuracy of the prediction. Then the combination model was applied to predict China＇s energy consumption in the next six years. By 2015,China＇s energy consumption will be 4.19 billion tons of standard coal.
出处
《科学技术与工程》
2010年第17期4267-4270,4282,共5页
Science Technology and Engineering
关键词
预测
神经网络
时间序列
组合模型
prediction neural network time series combination model
