国际石油价格时间序列的混沌分析与预测  

Using Chaos Theory Time-Series for Analysis and Prediction of Global Oil Price Patterns

作　　者：王洲[1] 马燕林[1] 

机构地区：[1]中央财经大学信息学院,北京100081

出　　处：《资源科学》2008年第12期1791-1795,共5页Resources Science

摘　　要：石油是当今世界最为重要的基础能源、化工原料和战略资源,它不仅支撑着石油生产国和消费国的经济发展,而且联系着各国国民经济的发展、人民生活水平的提高和国防安全。因此,石油价格是当前全球关心的普遍问题。研究表明,石油市场是具有混沌特性的复杂非线性系统。本文在非线性系统及复杂性理论框架内,采用相空间重构技术,提取描述吸引子特征量参数,定量的证明石油价格演化过程具有混沌特性,并采用混沌时间序列预测法预测石油价格走势。对近3年国际石油价格走势做了短期预测,预测结果表明,如无重大突发事件发生,2008年国际石油平均价格将有所上升,会在高价位持续震荡;2009年则会有所回落,将在每桶64美元上下波动。Oil is one of the most important fundamental global resources, as it supports the economic growth of both oil producing and consuming countries and is closely related to improved living standards and national defense. Oil prices are a pressing global issue, with major economic impacts. Oil markets are complex systems that are influenced by many factors in addition to the main factors of supply and demand. The mutual interactions among those factors make the markets highly complex and even a minor disturbance of one factor can cause market fluctuations. Experiments show that oil markets are nonlinear dynamic systems with chaotic features, and traditional statistical models don＇t accurately reflect the underlying dynamics. In order to mine the underlying features of oil markets and prices, this paper first applies a G-P algorithm to obtain embedded dimensions （m = 7）, which implies that the above-mentioned systems are multi-dimensional deterministic chaotic systems with fractal features. We then use the Phase Space Reconstruction Technique （PSRT） and Wolf algorithm to obtain the largest Lyapunov exponent （21） which describes the features of attractors. We found that the largest Lyapunov exponent is larger than zero （λ1 = 0. 117486782）, which proves that the evolution of oil prices has chaotic features. From the largest Lyapunov exponent, the maximum reliable prediction period is eight, which means the deviation will double if prediction time is more than eight years. Finally, the paper applies prediction methods of chaotic time-series to predict the trend of oil prices in the context of nonlinear systems and complexity science, using two methods to predict the future price： RBF neural network and forecasting based on maximum Lyapunov exponent. Results suggest that global oil prices will increase in 2008 and decline slightly in 2009. The paper concludes with a general discussion of oil price reforms in China.

关 键 词：石油价格 混沌理论 复杂性 预测 国际 

分 类 号：F416.22[经济管理—产业经济]