基于EEMD和神经网络的原油价格预测

发布时间：2018-02-20 18:37

本文关键词： 原油价格预测集合经验模式分解(EEMD) 神经网络模型　出处：《兰州大学》2017年硕士论文　论文类型：学位论文

【摘要】：国际原油价格的频繁波动很大程度上影响着全球经济和社会的稳定,而原油作为经济市场的一种特殊商品,其价格波动不仅受到供给与需求的影响,同时,天气状况、股票涨跌、经济发展、政治因素、国民心理预期等因素也对其有显著的影响。历史研究表明,原油价格的波动具有非线性、不确定性和动态性等特性,这些特性给原油价格的的建模与预测工作带来一定的挑战。因此,为提高模型的预测能力,在建模之前对数据去噪是很有必要的。本文在进行实证分析时,首先通过集合经验模式分解(EEMD)对国际原油价格加以处理得到一系列独立的固有模态函数(IMF)和残差序列,接着将前一步分解所得到的新序列分别作为BP神经网络模型的输入层,然后将各个BP神经网络预测的结果加总得到原油价格的最终预测值。为验证混合模型的有效性,本文对美国西德克萨斯中级原油价格(WTI)和布伦特原油价格(Brent)的数据进行预测,通过各项误差指标的对比,结果表明相比单个BP神经网络模型,本文提出的EEMD和神经网络模型对原油价格的预测更为准确。
[Abstract]:The frequent fluctuations in international crude oil prices to a large extent affect the stability of the global economy and society. As a special commodity in the economic market, the price fluctuations of crude oil are not only affected by supply and demand, but also by weather conditions. The rise and fall of stocks, economic development, political factors, national psychological expectations and other factors also have a significant impact on it. Historical studies show that the fluctuation of crude oil prices has the characteristics of nonlinearity, uncertainty and dynamics. These characteristics bring some challenges to the modeling and forecasting of crude oil price. Therefore, in order to improve the prediction ability of the model, it is necessary to de-noising the data before modeling. First, a series of independent intrinsic mode functions (IMF) and residual sequences are obtained by means of empirical mode decomposition (EEMD), and then the new sequences obtained by the former decomposition are used as input layers of BP neural network model. In order to verify the validity of the mixed model, the data of WTI and Brent crude prices of West Texas Intermediate crude are forecasted in this paper. The results show that the EEMD and neural network models proposed in this paper are more accurate than the single BP neural network model in predicting the price of crude oil.
【学位授予单位】：兰州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F416.22;F764.1;F224

【参考文献】