安徽大学学报（自然科学版）2025，Vol.49Issue(5)：1-10,10.DOI:10.3969/j.issn.1000-2162.2025.05.001
基于VMD-RNN-NM的农产品期货价格分解集成预测研究
Decomposition integration forecasting study of agricultural futures price based on VMD-RNN-NM
摘要
Abstract
In order to capture the complex fluctuation features in high-frequency data and improve the prediction accuracy of futures prices,this paper optimized the prediction model and adopted a decomposition integration strategy to construct a decomposition integration prediction model based on variational mode decomposition,recurrent neural network,and the downhill simplex method.Firstly,the original signal sequence was decomposed into multiple intrinsic model functions using variational mode decomposition.Then,each IMF value was predicted using an RNN combined with the grid search method.Finally,the optimal coefficients of the IMFs predicted values were determined using NM,and the final prediction results were obtained through weighted integration.To verify the effectiveness of the model,this study selected the 5 min trading price of agricultural futures as the research object.The empirical results demonstrated that the proposed decomposition integration prediction model significantly outperformed the single prediction model in terms of prediction accuracy.This finding indicated that by decomposing futures trading price data,the decomposition integration model effectively captured multi-scale features,thereby improving the prediction performance.Meanwhile,when aggregating the values of each IMF,this study assigned different coefficients to each IMF for weighted combination,which enhanced the model's accuracy compared with the traditional direct summing method.关键词
变分模态分解/循环神经网络/下山单纯形法/高频数据/分解集成预测Key words
variational mode decomposition/recurrent neural network/nelder-mead/high frequency data/decomposition integration forecasting分类
数理科学引用本文复制引用
袁宏俊,黄胜龙,胡凌云..基于VMD-RNN-NM的农产品期货价格分解集成预测研究[J].安徽大学学报（自然科学版）,2025,49(5):1-10,10.基金项目
国家自然科学基金资助项目(72371001) （72371001）
安徽省哲学社会科学规划项目(AHSKY2020D42) （AHSKY2020D42）
安徽省高校自然科学研究重点项目(2024AH050003,2022AH050602) （2024AH050003,2022AH050602）
