准确预测有色金属价格对于决策者、投资者和研究人员具有重要意义.为了提高预测精度,文中提出了一种新型混合预测模型,称为(EVMD-ICEEMDAN-RFEDformer,EIRF).首先,使用变分模态分解(variational mode decomposition,VMD)将原始价格分解为多个分量,同时使用改进的蚁狮搜索算法(modified ant lion optimization,MALO)对 VMD 的两个参数进行优化.其次,采用改进的带有自适应噪声的完全集成经验模式分解(improved complete ensemble empirical mode decomposition with adaptive noise,ICEEMDAN)进一步分解 VMD 产生的残差序列,从中提取有价值的信息.然后将所有分解的子序列输入到改进的频率增强分解变压器(reinforced frequency enhanced decomposition transformer,RFEDformer)中.最后,合并RFEDformer的预测并得出最终结果.为了验证模型的可靠性,文中利用了伦敦金属交易所的锡、铜和镍价格数据制定了 3个不同的实验,并与12个对比模型进行了比较.结果表明混合模型在3个数据集上都取得了良好的性能.
Nonferrous metal price prediction based on variational mode decomposition and reinforced frequency enhanced decomposition transformer
Accurately forecasting the prices of non-ferrous metals holds significant importance for decision-makers,investors,and researchers.To enhance prediction accuracy,this study introduces a novel hybrid prediction model called EIRF(EVMD-ICEEMDAN-RFEDformer).Initially,the original price undergoes decomposition into multiple components using variational mode decomposition(VMD),and the two parameters of VMD are optimized by modified ant lion search algorithm(MALO).Secondly,the improved fully integrated empirical mode decomposition with adaptive noise(ICEEMDAN)is applied to further break down the residual sequence produced by VMD,extracting valuable information from it.Finally,the forecasts from RFEDformer are combined to yield the final results.To validate the model's reliability,three experiments were conducted using price data for tin,copper,and nickel from the London Metal Exchange,comparing its performance with 12 benchmark models.The results indicate that the hybrid model demonstrates strong performance across all three datasets.
non-ferrous metal price forecastingant lion optimization algorithmquadratic decompositionRFEDformer modelSophia optimizerIKMSE loss function
万方数据
维普
