基于混合模型的国际原油价格预测研究

由于国际原油价格的剧烈波动，使得准确的原油价格预测极具挑战。为此，提出一种基于变分模态分解、季节性差分自回归滑动平均模型和果蝇优化最小二乘支持向量机的混合模型。利用变分模态分解方法将国际原油价格序列分解成一系列模态分量；针对周期性和非线性特征分量，分别建立季节性差分自回归滑动平均模型和果蝇优化最小二乘支持向量机模型进行预测；将各分量的预测值求和作为最终的预测结果。实证研究结果表明:所提混合模型相较对比模型能够明显提高国际原油价格的预测精度。

International Crude Oil Price Forecasting based on a Hybrid Model

Due to the serious fluctuations of international crude oil price, to accurately forecast crude oil price is very challenging. Therefore,a hybrid model based on variational mode decomposition(VMD),seasonal autoregressive integrated moving average (SARIMA)and least squares support vector machine(LSSVM)optimized by fruit fly optimization algorithm(FOA)was proposed. First, the international crude oil price series was decomposed into some mode components by VMD. Then,the SARIMA and LSSVM-FOA were established for periodic and nonlinear components, respectively. Finally, the summation of the forecast values of each component was used as the final forecasting result. Empirical results show that the proposed hybrid model can significantly improve the forecasting accuracy of international crude oil price compared with other models.