基于样本熵重构和PSO优化算法的人民币汇率预测  

RMB Exchange Rate Prediction Based on Sample Entropy Reconstruction and PSO Optimization Algorithm

作　　者：孙景云 赵盼盼 丁毅 SUN Jing-yun;ZHAO Pan-pan;DING Yi(School of Statistics,Lanzhou University of Finance and Economics,Lanzhou 730020,China)

机构地区：[1]兰州财经大学统计学院,甘肃兰州730020

出　　处：《数学的实践与认识》2021年第19期70-83,共14页Mathematics in Practice and Theory

基　　金：国家自然科学基金(72061020,71701084);甘肃省自然科学基金(21JRIRA280);甘肃省高等学校创新能力提升项目(2019A-060);兰州财经大学科研创新团队支持计划资助;兰州财经大学2020年度高等教育教学改革研究重点项目(LJZ202008)。

摘　　要：针对汇率数据具有随机性、非线性、高度波动性等复杂特征,提出了一种汇率预测的新方法(EEMD-SE-PSO-LSSVM).首先使用集合经验模态分解(EEMD)将原始美元兑人民币汇率序列分解为一系列的子序列.然后,通过样本熵(SE)量化各子序列的复杂度,将SE值接近的子序列进行合并重构.接着,通过相空间重构确定嵌入维数,并建立基于粒子群优化的最小二乘支持向量机(PSO-LSSVM)预测模型,然后对各子序列分别进行了预测.最后,对各子序列的预测结果进行非线性集成,得到汇率的最终预测值.通过对本文模型和其他多种模型预测结果的实证对比发现,采用本文提出的模型预测结果更准确,预测精度更高.Aiming at the complex characteristics of exchange rate data such as randomness,nonlinearity,and high volatility,this paper proposes a new method of exchange rate forecasting(EEMD-SE-PSO-LSSVM).First,the ensemble empirical mode decomposition(EEMD)is used to decompose the original USD/RMB exchange rate sequence into a series of sub-sequences.Second,the complexity of each subsequence is quantified by sample entropy(SE),and the subsequence with a smaller SE value is reconstructed.Then,the embedding dimension is determined through phase space reconstruction,and the least square support vector machine(PSO-LSSVM)prediction model based on particle swarm optimization is established,and then each sub-sequence is predicted separately.Finally,the non-linear integration of the prediction results of each subsequence is performed to obtain the final predicted value of the exchange rate.By comparing the prediction results of the model in this paper with other models,the empirical results show that the prediction results of the model proposed in this paper are more accurate and the prediction accuracy is higher.

关 键 词：汇率预测 集合经验模态分解 样本熵 粒子群优化 最小二乘支持向量机 

分 类 号：F832.6[经济管理—金融学] F224