基于模态分解与SRU网络的时间序列预测  

Time Series Forecasting Based on Modal Decomposition and SRU Network

作　　者：钱钧[1] 何曦[1] 冯焱侠 李维勤[2] QIAN Jun;HE Xi;FENG Yan-xia;LI Wei-qin(Xi'an Research Institute of Applied Optics,Xi'an 710065 China;School of Automation and Information Engineering,Xi'an University of technology,Xi'an 710048 China)

机构地区：[1]西安应用光学研究所,陕西西安710065 [2]西安理工大学自动化与信息工程学院,陕西西安710048

出　　处：《自动化技术与应用》2024年第8期99-104,共6页Techniques of Automation and Applications

基　　金：国家自然科学基金面上项目(62003259)资助。

摘　　要：时间序列预测在工业、农业、金融及军事等领域中具有重要的应用价值。为了进一步提高预测的可靠性和准确性,构建一种基于模态分解与SRU网络的杂交预测模型。首先,针对模态个数难以确定的问题,构建基于平均样本熵来确定模态个数的自适应变分模态分解(AVMD)模型,以减少不同频率上的混叠及降低随机噪声的干扰。通过在Adam算法中引入了随机调整参数,来提高SRU网络的训练速度及增强网络跳出局部最优解的能力。最后,发展一种基于AVMD与SRU网络的杂交模型。为评估提出的预测模型的可靠性和准确性,将之与一些最新预测方法做比较。电力负荷序列的实验结果表明,所提出的杂交预测模型具有较高的准确性和可靠性。Time series forecasting has important application value in the fields of industry,agriculture,finance and military.In order to further improve the reliability and accuracy of the forecasting result,a hybrid model based on modal decomposition and SRU network is established in this paper.Firstly,aiming at the problem that it is difficult to determine the number of modes in variational modal decomposition,the average sample entropy is developed to determine the number of modes,and an adaptive variational modal decomposition(AVMD)model is proposed to reduce aliasing and sequence randomness and improve the robustness of the forecasting model.Moreover,the random adjustment parameters are introduced into the Adam algorithm of the simple recursive unit(SRU)network to improve the training speed and the ability to jump out of local optimization.Finally,a hybrid forecasting model of AVMD and SRU network is proposed.In order to verify the reliability and accuracy of the proposed hybrid model,it is compared with some state of art methods.Results of the actual load series show that the forecasting model developed in this paper has high accuracy and reliability.

关 键 词：预测 时间序列 模态分解 平均样本熵 随机调整参数 循环单元 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP393.09[自动化与计算机技术—控制科学与工程] TM715[电气工程—电力系统及自动化]