Researches on Time Series Prediction with Echo State Networks
电子学报2010年38卷第2A期 页码:148-154
作者机构:
1. 1.哈尔滨工业大学自动化测试与控制研究所,黑龙江,哈尔滨,150001
2. 哈尔滨理工大学电子科学与技术系,黑龙江,哈尔滨,150080
作者简介:
基金信息:
DOI:
中图分类号:TP183
纸质出版:2010
稿件说明:
移动端阅览
FONT face, Verdana, 彭 宇, 等. 基于回声状态网络的时间序列预测方法研究[J]. 电子学报, 2010,38(2A):148-154.
FONT face, Verdana, PENG Yu, et al. Researches on Time Series Prediction with Echo State Networks[J]. Acta Electronica Sinica, 2010, 38(2A): 148-154.
FONT face, Verdana, 彭 宇, 等. 基于回声状态网络的时间序列预测方法研究[J]. 电子学报, 2010,38(2A):148-154.DOI:
FONT face, Verdana, PENG Yu, et al. Researches on Time Series Prediction with Echo State Networks[J]. Acta Electronica Sinica, 2010, 38(2A): 148-154.DOI:
<FONT face=Verdana>针对回声状态网络(Echo State Networks,ESNs)输入序列延迟时间(和嵌入维数D的选择以及储备池的适应性问题,利用自相关性分析法从被预测样本序列构建ESNs网络输入,并通过移动通信话务量的预测问题,采用实验分析的方法讨论了储备池参数选择对于时间序列预测性能的影响.与采用ARMA和BP神经网络的预测方法相比,新方法在保证预测精度和效率的情况下,具有更好的泛化能力.
Abstract
<FONT face=Verdana>The choice of delay time and embedded dimension in time series modeling and prediction and the problem of reservoir adaption are challenges for Echo State Networks (ESNs).Correlation analysis is introduced to construct the inputs vector from the time series in ESNs networks.Moreover
the effects of different parameters settings on prediction performances in reservoir is analyzed by experiments of mobile communication traffic prediction.Compared with ARMA and BP neural networks
the proposed method can ensure not only the accuracy and efficiency but also the good generalities.