基于CPSO - SVM的石油需求预测

摘要

In order to accurately forecast petroleum demand, a petroleum demand forecasting method was presented based on chaos particle swarm optimization ans support vector machine ( CPSO - SVM). The CPSO was used for SVM parameter optimization, and then SVM was used to buid the model of nonlinear variation rule of petroleum demand, Finally, the petroleum demand data from 1989 to 2007 were used for simulation. The results show that, compared with other petroleum demand forecast algorithmd, the CPSO - SVM raises oil demand forecast accuracy and provides a new method for predicting of petroleum demand.%研究石油需求准确预测问题,石油是一种国际货物,受多种因素影响,具有不确定性、非线性和突变性等特点,传统线性建模方法预测精度低.为了提高石油需求预测精度,提出一种利用混沌粒子群优化支持向量机的非线性石油需求预测方法( CPSO - SVM).采用CPSO优化SVM参数,然后采用优化的SVM对石油需求的非线性变化规律进行建模,以1989～2007年石油需求数据进行仿真,结果表明,采用CPSO - SVM提高了石油需求量的预测精度,为石油需求量预测提供一种新的研究方法.