摘要
为了准确把握全球稀土消费变化,本文在分析单整自回归移动平均(ARIMA)模型与非线性灰色伯努利(NGBM)模型特点的基础上,采用粒子群优化算法(PSO)对非线性灰色伯努利模型的参数进行了优选,建立了ARIMA耦合PSO-NGBM的全球稀土消费的时间序列预测模型。该模型将全球稀土消费时间序列的数据结构分解为线性自相关主体和非线性残差两部分,首先用ARIMA模型预测序列的线性主体,然后用PSO-NGBM模型对其非线性残差进行估计,最终合成为整个序列的预测结果。预测结果表明,耦合模型的预测准确率显著高于单一的ARIMA模型的预测准确率,从而证实了耦合模型用于全球稀土消费预测的有效性。
In order to accurately understand the global rare earth consumption trend, an integrated prediction model combining ARI- MA and PSO- NGBM was established based on the analysis of Auto- regressive Integrated Moving Average （ARIMA） and Nonlinear Gray Bernoulli model （NGBM）. The parameters of Nonlinear Gray Bernoulli model （NGBM） was optimized with a particle swarm optimization （PSO） algorithm. In the integrated prediction model, the structure of world rare earth consumption time series was divided into two parts including a linear autocorrelation structure and nonlinear structure. AR/MA was used to predict the linear component of global rare earth consumption time series, and the NGBM model was applied to the nonlinear residuals component. The results of global rare earth consump- tion prediction showed that the integrated model had more accuracy for predicting global rare earth consumption than that of single model. The results implied that the integrated model can be used to predict effectively the future consumption of global rare earth.
出处
《工业技术经济》
CSSCI
北大核心
2013年第7期110-116,共7页
Journal of Industrial Technological Economics
基金
国土资源部公益性行业科研专项课题"限制
允许和鼓励开采固体非能源矿产资源分类标准研究"(项目编号:201211067-3)
