基于分解-集成混合算法的原油价格预测

Crude Oil Price Forecasting Based on Decomposition-Integration Hybrid Algorithm

原油作为国际市场上的一种特殊商品,其价格的波动会在很大程度上影响世界经济的可持续发展。由于原油价格具有强烈的非线性性和波动性,本文提出了一种新的混合算法来对原油价格进行有效预测。首先利用集成经验模态分解算法对原油价格进行分解,其次采用混沌粒子群算法优化极限学习机的神经元个数,利用优化后的极限学习机对分解出的子序列进行预测,最后利用集成方法得到原油价格的最终预测值。为了检验混合算法在原油价格预测中的优越性和适用性,本文对美国西德克萨斯中级原油价格和布伦特原油价格分别进行预测,并从统计评价指标、统计检验以及鲁棒性三个方面,验证了混合模型在原油价格的预测中具有更好的预测性能。

As a special commodity in the international market,crude oil＇s price fluctuations will,to a large extent,affect the stable development of the world economy. Due to the strong nonlinearity and volatility of crude oil prices,a new hybrid algorithm is proposed to predict crude oil prices. Firstly,the crude oil price is decomposed by the integrated empirical mode decomposition algorithm. Secondly,the number of neurons of the extreme learning machine is optimized by the chaotic particle swarm optimization algorithm. The optimized sub-sequence is predicted by the optimized limit learning machine,and finally the integration method is used. Get the final price of crude oil. In order to test the superiority and applicability of the hybrid algorithm in crude oil price forecasting,this paper separately forecasts the data of the West Texas Intermediate crude oil price and the Brent crude oil price in the United States,and estimates the statistical indicators,statistical tests and robust analysis of the three aspects verified that the hybrid model has better predictive performance.