搜索

x

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于Nyström柯西核共轭梯度算法的混沌时间序列预测

齐乐天 王世元 沈明琳 黄刚毅

引用本文:
Citation:

基于Nyström柯西核共轭梯度算法的混沌时间序列预测

齐乐天, 王世元, 沈明琳, 黄刚毅

Prediction of chaotic time series based on Nyström Cauchy kernel conjugate gradient algorithm

Qi Le-Tian, Wang Shi-Yuan, Shen Ming-Lin, Huang Gang-Yi
PDF
HTML
导出引用
  • 混沌时间序列能够较好反映真实环境的非线性和非平稳性特性, 然而具有二阶统计特性的核自适应滤波器(kernel adaptive filter, KAF)在处理含噪声和异常值的混沌时间序列时, 其预测性能显著下降. 为提高核自适应滤波器的鲁棒性, 本文提出了一种用于测量非线性相似度的柯西核损失(Cauchy kernel loss, CKL), 并采用半平方(half-quadratic, HQ)方法保证了CKL的全局凸性. 为改善随机梯度下降法收敛速度较慢且容易陷入局部最优的不足, 采用共轭梯度(conjugate gradient, CG)方法优化CKL. 进一步, 为解决核矩阵网络增长的问题, 采取Nyström稀疏策略近似核矩阵, 并利用概率密度秩量化(probability density rank-based quantization, PRQ)提高逼近精度. 基于此, 本文提出了一种新的基于Nyström和PRQ的柯西核共轭梯度(Nyström Cauchy kernel conjugate gradient with PRQ, NCKCG-PRQ)算法有效实现了混沌时间序列的预测. 基于合成和真实两类混沌时间序列验证了所提NCKCG-PRQ算法在稳态性能, 鲁棒性和计算存储复杂度上的优势.
    Chaotic time series can well reflect the nonlinearity and non-stationarity of real environment changes. The traditional kernel adaptive filter (KAF) with second-order statistical characteristics suffers performance degeneration dramatically for predicting chaotic time series containing noises and outliers. In order to improve the robustness of adaptive filters in the presence of impulsive noise, a nonlinear similarity measure named Cauchy kernel loss (CKL) is proposed, and the global convexity of CKL is guaranteed by the half-quadratic (HQ) method. To improve the convergence rate of stochastic gradient descent and avoid a local optimum simultaneously, the conjugate gradient (CG) method is used to optimize CKL. Furthermore, to address the issue of kernel matrix network growth, the Nyström sparse strategy is adopted to approximate the kernel matrix and then the probability density rank-based quantization (PRQ) is used to improve the approximation accuracy. To this end, a novel Nyström Cauchy kernel conjugate gradient with PRQ (NCKCG-PRQ) algorithm is proposed for the prediction of chaotic time series in this paper. Simulations on prediction of synthetic and real-world chaotic time series validate the advantages of the proposed algorithm in terms of filtering accuracy, robustness, and computational storage complexity.
      通信作者: 王世元, wsy@swu.edu.cn
    • 基金项目: 国家自然科学基金(批准号: 62071391)、重庆市自然科学基金面上项目(批准号: cstc2020jcyj-msxmX0234)和中央高校基本科研业务费(批准号: 2020jd001)资助的课题.
      Corresponding author: Wang Shi-Yuan, wsy@swu.edu.cn
    • Funds: Project supported by the National Nature Science Foundation of China (Grant No. 62071391), Natural Science Foundation of Chongqing (Grant No. cstc2020jcyj-msxmX0234), and Fundamental Research Funds for the Central Universities (Grant No. 2020jd001).
    [1]

    林毅, 刘文波, 沈骞 2018 物理学报 67 230502Google Scholar

    Lin Y, Liu W B, Shen Q 2018 Acta Phys. Sin. 67 230502Google Scholar

    [2]

    王梦蛟, 吴中堂, 冯久超 2015 物理学报 64 040503Google Scholar

    Wang M J, Wu Z T, Feng J C 2015 Acta Phys. Sin. 64 040503Google Scholar

    [3]

    唐舟进, 任峰, 彭涛, 王文博 2014 物理学报 63 050505Google Scholar

    Tang Z J, Ren F, Peng T, Wang W B 2014 Acta Phys. Sin. 63 050505Google Scholar

    [4]

    梅英, 谭冠政, 刘振焘, 武鹤 2018 物理学报 67 080502Google Scholar

    Mei Y, Tan G Z, Liu Z T, Wu H 2018 Acta Phys. Sin. 67 080502Google Scholar

    [5]

    王新迎, 韩敏, 王亚楠 2013 物理学报 62 050504Google Scholar

    Wang X Y, Han M, Wang Y N 2013 Acta Phys. Sin. 62 050504Google Scholar

    [6]

    王世元, 史春芬, 钱国兵, 王万里 2018 物理学报 67 018401Google Scholar

    Wang S Y, Shi C F, Qian G B, Wang W L 2018 Acta Phys. Sin. 67 018401Google Scholar

    [7]

    Peng L B, Li X F, Bi D J, Xie Y L 2018 Signal Process. Lett. 25 1335Google Scholar

    [8]

    赵永平, 张丽艳, 李德才, 王立峰, 蒋洪章 2013 物理学报 62 120511Google Scholar

    Zhao Y P, Zhang L Y, Li D C, Wang L F, Jiang H Z 2013 Acta Phys. Sin. 62 120511Google Scholar

    [9]

    张家树, 党建亮, 李恒超 2007 物理学报 56 67Google Scholar

    Zhang J S, Dang J L, Li H C 2007 Acta Phys. Sin. 56 67Google Scholar

    [10]

    张洪宾, 孙小端, 贺玉龙 2014 物理学报 63 040505Google Scholar

    Zhang H B, Sun X D, He Y L 2014 Acta Phys. Sin. 63 040505Google Scholar

    [11]

    唐舟进, 彭涛, 王文博 2014 物理学报 63 130504Google Scholar

    Tang Z J, Peng T, Wang W B 2014 Acta Phys. Sin. 63 130504Google Scholar

    [12]

    火元莲, 王丹凤, 龙小强, 连培君, 齐永锋 2021 物理学报 70 158401Google Scholar

    Huo Y L, Wang D F, Long X Q, Lian P J, Qi Y F 2021 Acta Phys. Sin. 70 158401Google Scholar

    [13]

    火元莲, 王丹凤, 龙小强, 连培君, 齐永锋 2021 物理学报 70 028401Google Scholar

    Huo Y L, Wang D F, Long X Q, Lian P J, Qi Y F 2021 Acta Phys. Sin. 70 028401Google Scholar

    [14]

    Wu Z, Shi J, Xie Z, Ma W 2015 Signal Process. 117 11Google Scholar

    [15]

    Liu W F, Pokharel P P, Príncipe J C 2008 IEEE Trans. Signal Process. 56 543Google Scholar

    [16]

    Engel Y, Mannor S, Meir R 2004 IEEE Trans. Signal Process. 52 2275Google Scholar

    [17]

    Chen B D, Príncipe J C 2012 Signal Process. Lett. 19 491Google Scholar

    [18]

    Li C G, Shen P C, Liu Y, Zhang Z Y 2013 IEEE Trans. Signal Process. 61 4011Google Scholar

    [19]

    Chen B D, Xing L, Zhao H Q, Zheng N N, Príncipe J C 2016 IEEE Trans. Signal Process. 64 3376Google Scholar

    [20]

    Li X L, Lu Q M, Dong Y S, Tao D C 2019 Trans. Neural Netw. Learn. Syst. 30 2067Google Scholar

    [21]

    Shi W, Xiong K, Wang S Y 2019 IEEE Access. 7 120548Google Scholar

    [22]

    Lei Y W, Hu T, Li G Y, Tang K 2020 Trans. Neural Netw. Learn. Syst. 31 4394Google Scholar

    [23]

    Zhang M, Wang X J, Chen X M, Zhang A X 2018 IEEE Trans. Signal Process. 66 4377Google Scholar

    [24]

    Xiong K, Herbert H C, Wang S Y 2021 IEEE Trans. Cybern. 51 5497Google Scholar

    [25]

    Chen B D, Zhao S L, Zhu P P, Príncipe J C 2013 Trans. Neural Netw. Learn. Syst. 24 1484Google Scholar

    [26]

    Zhang T, Wang S Y, Huang X W, Jia L 2020 Signal Process. Lett. 27 361Google Scholar

    [27]

    Zhang T, He F L, Zheng Z, Wang S Y 2020 IEEE Trans. Circuits Syst. Express Briefs 67 2772Google Scholar

    [28]

    He F L, Xiong K, Wang S Y 2020 IEEE Access. 8 18716Google Scholar

    [29]

    Zhang T, Wang S Y 2020 Signal Process. Lett. 27 1535Google Scholar

    [30]

    Qi L T, Shen M L, Wang D L, Wang S Y 2021 Signal Process. Lett. 28 1011Google Scholar

    [31]

    Zhang H N, Yang B, Wang L, Wang S Y 2021 IEEE Trans. Signal Process. 69 1859Google Scholar

    [32]

    Xiong K, Wang S Y 2019 Signal Process. Lett. 26 740Google Scholar

    [33]

    Qin Z D, Chen B D, GU Y T, Zheng N N, Príncipe J C 2020 IEEE Trans. Neural Netw. Learn. Syst. 31 3100Google Scholar

    [34]

    Zheng Y F, Wang S Y, Feng J C, Tse C K 2016 Digit. Signal Process. 48 130Google Scholar

    [35]

    Weng B W, Barner K E 2005 IEEE Trans. Signal Process. 53 2588Google Scholar

    [36]

    Huang X W, Wang S Y, Xiong K 2019 Symmetry 11 1323Google Scholar

  • 图 1  示波器示意图

    Fig. 1.  Schematic diagram of oscilloscope.

    图 2  蔡氏电路原理图

    Fig. 2.  Schematic diagram of the Chua’s circuit.

    图 3  不同采样点个数$ m $对NCKCG-PRQ算法的稳态MSE值和平均运算时间的影响 (a) MG混沌时间序列; (b) 蔡氏混沌时间序列

    Fig. 3.  Influence of different number of sampling points on steady-state MSE value and average operation time of NCKCG-PRQ algorithm: (a) MG chaotic time series; (b) chaotic time series based on Chua’s circuit.

    图 4  在脉冲噪声环境下不同算法的测试MSE学习曲线 (a) MG时间序列 ; (b)蔡氏混沌时间序列

    Fig. 4.  Testing MSE learning curves of different algorithms in impulsive noise environment: (a) MG chaotic time series; (b) chaotic time series based on Chua’s circuit.

    图 5  NCKCG-PRQ算法对测试数据的最终预测结果 (a) MG混沌时间序列; (b) 蔡氏混沌时间序列

    Fig. 5.  Final predicted results of NCKCG-PRQ algorithm for the test sets: (a) MG chaotic time series; (b) chaotic time series based on Chua’s circuit.

    表 1  NCKCG-PRQ算法

    Table 1.  NCKCG-PRQ algorithm.

      输入: 输入输出对$\left\{ { {{\boldsymbol{u}}_k}, {d_k} } \right\}, k{\text{ = } }1, 2, \cdot \cdot \cdot$
      初始化参数: $ {\boldsymbol{\hat u}}(i) $为PRQ采样后样本点; $ {\boldsymbol{\varLambda }} $和W分别为有关$ {\boldsymbol{\hat U}} $核矩阵的特征值降序排列的对角矩阵和对应特征向量构成列正交矩阵; ${ {\boldsymbol{K} }_c}(1) = [\kappa ({\boldsymbol{u} }(1), {\boldsymbol{\hat u} }(1)), \cdots , \kappa ({\boldsymbol{u} }(1), {\boldsymbol{\hat u} }(m))]$为初始核向量; 映射输入$ {{\boldsymbol{z}}_1} = {{\boldsymbol{\varLambda }}^{ - 1/2}}{{\boldsymbol{W}}^{\text{T}}}{{\boldsymbol{K}}_c}{(1)^{\text{T}}} $; 权重$ {\boldsymbol{\varOmega }}_1^z = 0 $; 加权函数$ {\zeta _1} = {\exp}( - {s_1})/\left\{ {{\delta ^2}\left[ {{\eta ^{ - 1}} + {s_1}{\exp}( - {s_1})} \right]} \right\} $; 期望$ {d_1}{\text{ = }}{({{\boldsymbol{\varOmega }}_1}^z)^{\text{T}}}{{\boldsymbol{z}}_1} $; 相关矩阵$ R_1^z = {\zeta _1}{{\boldsymbol{z}}_1}{\boldsymbol{z}}_1^{\rm T} $; 互相关向量${\boldsymbol{c}}_1^z={\zeta _1}{d_1}{{\bf{z}}_1}$; 冗余向量${\boldsymbol{r}}_1^z ={\boldsymbol{ c}}_1^z - {\boldsymbol{R}}_1^z{\boldsymbol{\varOmega } }_1^z$; 方向向量${\boldsymbol{p}}_1^z = {\boldsymbol{r}}_1^z$, 遗忘因子$ \lambda {\text{ = }}0.999 $
      循环$ \left( {k{\text{ = 2}}, 3, \cdot \cdot \cdot } \right) $:
      1.输入核向量${ {\boldsymbol{K} }_c}(i) = [\kappa ({\boldsymbol{u} }(i), {\boldsymbol{\hat u} }(1)), \cdots , \kappa ({\boldsymbol{u} }(i), {\boldsymbol{\hat u} }(m)]$;
      2.映射输入${\boldsymbol{z} }( \cdot ) = { {\boldsymbol{\varLambda } }^{ - 1/2} }{ {\boldsymbol{W} }^{\text{T} } }{[\kappa ( \cdot , {\boldsymbol{\hat u} }(1)), \cdots , \kappa ( \cdot , {\boldsymbol{\hat u} }(m))]^{\text{T} } }$;
      3.误差更新$ {e_{k + 1}} = {d_{k + 1}} - {\left( {{\boldsymbol{\varOmega }}_k^z} \right)^{\text{T}}}{{\boldsymbol{z}}_{k + 1}} $, 加权函数$ {\zeta _k} = {\exp}( - {s_k})/\left\{ {{\delta ^2}\left[ {{\eta ^{ - 1}} + {s_k}{\exp}( - {s_k})} \right]} \right\} $
      4.自相关矩阵更新${\boldsymbol{R}}_{k + 1}^z = \lambda {\boldsymbol{R}}_k^z + {\zeta _{k + 1} }{ {\boldsymbol{z} }_{k + 1} }{\boldsymbol{z} }_{k + 1}^{\rm T}$, 计算步长${\alpha _k} = \frac{ { { {\left( {{\boldsymbol{p}}_k^z} \right)}^{\text{T} } }{\boldsymbol{r}}_k^z} }{ { { {\left( {{\boldsymbol{p}}_k^z} \right)}^{\text{T} } }{\boldsymbol{R}}_{k + 1}^z{\boldsymbol{p}}_k^z} }$;
      5.权重更新${\boldsymbol{\varOmega } }_{k + 1}^z = {\boldsymbol{\varOmega } }_k^z + {\alpha _k}{\boldsymbol{p}}_k^z$, 残差向量更新${\boldsymbol{r}}_{k + 1}^z = \lambda {\boldsymbol{r}}_k^z - {\alpha _k}{\boldsymbol{R}}_{k + 1}^z{\boldsymbol{p}}_k^z + {\zeta _{k + 1} }{{\bf{z}}_{k + 1} }{{\boldsymbol{e}}_{k + 1} }$;
      6.计算步长 ${\beta _k} = \frac{ { { {\left( {{\boldsymbol{r}}_{k + 1}^z} \right)}^{\text{T} } }\left( {{\boldsymbol{r}}_{k + 1}^z - {\boldsymbol{r}}_k^z} \right)} }{ { { {\left( {{\boldsymbol{r}}_k^z} \right)}^{\text{T} } }{\boldsymbol{r}}_k^z} }$, 共轭方向更新${\boldsymbol{p}}_{k + 1}^z = {\boldsymbol{r}}_{k + 1}^z + {\beta _k}r_k^z$
      循环终止
    下载: 导出CSV

    表 2  不同算法在MG混沌时间序列中的仿真结果

    Table 2.  Simulation results of different algorithms in MG chaotic time series.

    实验模型算法字典数目运算时间/s稳态MSE/dB
    MG混沌时间序列CKCG200041.486–35.443
    RFFCCG602.095–32.006
    NKRGMC-PRQ602.383–33.068
    NKCG-KM603.745N/A
    NCKCG-PRQ601.584–34.446
    下载: 导出CSV

    表 3  不同算法在蔡氏电路混沌时间序列中的仿真结果

    Table 3.  Simulation results of different algorithms in chaotic time series based on Chua's circuit.

    实验模型算法字典数目运算时间/s稳态MSE/dB
    蔡氏混沌时间序列CKCG200042.618–35.840
    RFFCCG131.128–34.926
    NKRGMC-PRQ130.988–35.819
    NKCG-KM131.146N/A
    NCKCG-PRQ130.957–35.865
    下载: 导出CSV
  • [1]

    林毅, 刘文波, 沈骞 2018 物理学报 67 230502Google Scholar

    Lin Y, Liu W B, Shen Q 2018 Acta Phys. Sin. 67 230502Google Scholar

    [2]

    王梦蛟, 吴中堂, 冯久超 2015 物理学报 64 040503Google Scholar

    Wang M J, Wu Z T, Feng J C 2015 Acta Phys. Sin. 64 040503Google Scholar

    [3]

    唐舟进, 任峰, 彭涛, 王文博 2014 物理学报 63 050505Google Scholar

    Tang Z J, Ren F, Peng T, Wang W B 2014 Acta Phys. Sin. 63 050505Google Scholar

    [4]

    梅英, 谭冠政, 刘振焘, 武鹤 2018 物理学报 67 080502Google Scholar

    Mei Y, Tan G Z, Liu Z T, Wu H 2018 Acta Phys. Sin. 67 080502Google Scholar

    [5]

    王新迎, 韩敏, 王亚楠 2013 物理学报 62 050504Google Scholar

    Wang X Y, Han M, Wang Y N 2013 Acta Phys. Sin. 62 050504Google Scholar

    [6]

    王世元, 史春芬, 钱国兵, 王万里 2018 物理学报 67 018401Google Scholar

    Wang S Y, Shi C F, Qian G B, Wang W L 2018 Acta Phys. Sin. 67 018401Google Scholar

    [7]

    Peng L B, Li X F, Bi D J, Xie Y L 2018 Signal Process. Lett. 25 1335Google Scholar

    [8]

    赵永平, 张丽艳, 李德才, 王立峰, 蒋洪章 2013 物理学报 62 120511Google Scholar

    Zhao Y P, Zhang L Y, Li D C, Wang L F, Jiang H Z 2013 Acta Phys. Sin. 62 120511Google Scholar

    [9]

    张家树, 党建亮, 李恒超 2007 物理学报 56 67Google Scholar

    Zhang J S, Dang J L, Li H C 2007 Acta Phys. Sin. 56 67Google Scholar

    [10]

    张洪宾, 孙小端, 贺玉龙 2014 物理学报 63 040505Google Scholar

    Zhang H B, Sun X D, He Y L 2014 Acta Phys. Sin. 63 040505Google Scholar

    [11]

    唐舟进, 彭涛, 王文博 2014 物理学报 63 130504Google Scholar

    Tang Z J, Peng T, Wang W B 2014 Acta Phys. Sin. 63 130504Google Scholar

    [12]

    火元莲, 王丹凤, 龙小强, 连培君, 齐永锋 2021 物理学报 70 158401Google Scholar

    Huo Y L, Wang D F, Long X Q, Lian P J, Qi Y F 2021 Acta Phys. Sin. 70 158401Google Scholar

    [13]

    火元莲, 王丹凤, 龙小强, 连培君, 齐永锋 2021 物理学报 70 028401Google Scholar

    Huo Y L, Wang D F, Long X Q, Lian P J, Qi Y F 2021 Acta Phys. Sin. 70 028401Google Scholar

    [14]

    Wu Z, Shi J, Xie Z, Ma W 2015 Signal Process. 117 11Google Scholar

    [15]

    Liu W F, Pokharel P P, Príncipe J C 2008 IEEE Trans. Signal Process. 56 543Google Scholar

    [16]

    Engel Y, Mannor S, Meir R 2004 IEEE Trans. Signal Process. 52 2275Google Scholar

    [17]

    Chen B D, Príncipe J C 2012 Signal Process. Lett. 19 491Google Scholar

    [18]

    Li C G, Shen P C, Liu Y, Zhang Z Y 2013 IEEE Trans. Signal Process. 61 4011Google Scholar

    [19]

    Chen B D, Xing L, Zhao H Q, Zheng N N, Príncipe J C 2016 IEEE Trans. Signal Process. 64 3376Google Scholar

    [20]

    Li X L, Lu Q M, Dong Y S, Tao D C 2019 Trans. Neural Netw. Learn. Syst. 30 2067Google Scholar

    [21]

    Shi W, Xiong K, Wang S Y 2019 IEEE Access. 7 120548Google Scholar

    [22]

    Lei Y W, Hu T, Li G Y, Tang K 2020 Trans. Neural Netw. Learn. Syst. 31 4394Google Scholar

    [23]

    Zhang M, Wang X J, Chen X M, Zhang A X 2018 IEEE Trans. Signal Process. 66 4377Google Scholar

    [24]

    Xiong K, Herbert H C, Wang S Y 2021 IEEE Trans. Cybern. 51 5497Google Scholar

    [25]

    Chen B D, Zhao S L, Zhu P P, Príncipe J C 2013 Trans. Neural Netw. Learn. Syst. 24 1484Google Scholar

    [26]

    Zhang T, Wang S Y, Huang X W, Jia L 2020 Signal Process. Lett. 27 361Google Scholar

    [27]

    Zhang T, He F L, Zheng Z, Wang S Y 2020 IEEE Trans. Circuits Syst. Express Briefs 67 2772Google Scholar

    [28]

    He F L, Xiong K, Wang S Y 2020 IEEE Access. 8 18716Google Scholar

    [29]

    Zhang T, Wang S Y 2020 Signal Process. Lett. 27 1535Google Scholar

    [30]

    Qi L T, Shen M L, Wang D L, Wang S Y 2021 Signal Process. Lett. 28 1011Google Scholar

    [31]

    Zhang H N, Yang B, Wang L, Wang S Y 2021 IEEE Trans. Signal Process. 69 1859Google Scholar

    [32]

    Xiong K, Wang S Y 2019 Signal Process. Lett. 26 740Google Scholar

    [33]

    Qin Z D, Chen B D, GU Y T, Zheng N N, Príncipe J C 2020 IEEE Trans. Neural Netw. Learn. Syst. 31 3100Google Scholar

    [34]

    Zheng Y F, Wang S Y, Feng J C, Tse C K 2016 Digit. Signal Process. 48 130Google Scholar

    [35]

    Weng B W, Barner K E 2005 IEEE Trans. Signal Process. 53 2588Google Scholar

    [36]

    Huang X W, Wang S Y, Xiong K 2019 Symmetry 11 1323Google Scholar

  • [1] 黄颖, 顾长贵, 杨会杰. 神经网络超参数优化的删除垃圾神经元策略. 物理学报, 2022, 71(16): 160501. doi: 10.7498/aps.71.20220436
    [2] 王世元, 史春芬, 钱国兵, 王万里. 基于分数阶最大相关熵算法的混沌时间序列预测. 物理学报, 2018, 67(1): 018401. doi: 10.7498/aps.67.20171803
    [3] 田中大, 高宪文, 石彤. 用于混沌时间序列预测的组合核函数最小二乘支持向量机. 物理学报, 2014, 63(16): 160508. doi: 10.7498/aps.63.160508
    [4] 唐舟进, 彭涛, 王文博. 一种基于相关分析的局域最小二乘支持向量机小尺度网络流量预测算法. 物理学报, 2014, 63(13): 130504. doi: 10.7498/aps.63.130504
    [5] 唐舟进, 任峰, 彭涛, 王文博. 基于迭代误差补偿的混沌时间序列最小二乘支持向量机预测算法. 物理学报, 2014, 63(5): 050505. doi: 10.7498/aps.63.050505
    [6] 王新迎, 韩敏, 王亚楠. 含噪混沌时间序列预测误差分析. 物理学报, 2013, 62(5): 050504. doi: 10.7498/aps.62.050504
    [7] 王新迎, 韩敏. 基于极端学习机的多变量混沌时间序列预测. 物理学报, 2012, 61(8): 080507. doi: 10.7498/aps.61.080507
    [8] 宋彤, 李菡. 基于小波回声状态网络的混沌时间序列预测. 物理学报, 2012, 61(8): 080506. doi: 10.7498/aps.61.080506
    [9] 宋青松, 冯祖仁, 李人厚. 用于混沌时间序列预测的多簇回响状态网络. 物理学报, 2009, 58(7): 5057-5064. doi: 10.7498/aps.58.5057
    [10] 张勇, 关伟. 基于最大Lyapunov指数的多变量混沌时间序列预测. 物理学报, 2009, 58(2): 756-763. doi: 10.7498/aps.58.756
    [11] 杨永锋, 任兴民, 秦卫阳, 吴亚锋, 支希哲. 基于EMD方法的混沌时间序列预测. 物理学报, 2008, 57(10): 6139-6144. doi: 10.7498/aps.57.6139
    [12] 张军峰, 胡寿松. 基于多重核学习支持向量回归的混沌时间序列预测. 物理学报, 2008, 57(5): 2708-2713. doi: 10.7498/aps.57.2708
    [13] 韩 敏, 史志伟, 郭 伟. 储备池状态空间重构与混沌时间序列预测. 物理学报, 2007, 56(1): 43-50. doi: 10.7498/aps.56.43
    [14] 孟庆芳, 张 强, 牟文英. 混沌时间序列多步自适应预测方法. 物理学报, 2006, 55(4): 1666-1671. doi: 10.7498/aps.55.1666
    [15] 胡玉霞, 高金峰. 一种预测混沌时间序列的模糊神经网络方法. 物理学报, 2005, 54(11): 5034-5038. doi: 10.7498/aps.54.5034
    [16] 甘建超, 肖先赐. 基于相空间邻域的混沌时间序列自适应预测滤波器(Ⅰ)线性自适应滤波. 物理学报, 2003, 52(5): 1096-1101. doi: 10.7498/aps.52.1096
    [17] 甘建超, 肖先赐. 基于相空间邻域的混沌时间序列自适应预测滤波器(Ⅱ)非线性自适应滤波. 物理学报, 2003, 52(5): 1102-1107. doi: 10.7498/aps.52.1102
    [18] 韦保林, 罗晓曙, 汪秉宏, 全宏俊, 郭维, 傅金阶. 一种基于三阶Volterra滤波器的混沌时间序列自适应预测方法. 物理学报, 2002, 51(10): 2205-2210. doi: 10.7498/aps.51.2205
    [19] 张家树, 肖先赐. 混沌时间序列的Volterra自适应预测. 物理学报, 2000, 49(3): 403-408. doi: 10.7498/aps.49.403
    [20] 张家树, 肖先赐. 混沌时间序列的自适应高阶非线性滤波预测. 物理学报, 2000, 49(7): 1221-1227. doi: 10.7498/aps.49.1221
计量
  • 文章访问数:  4801
  • PDF下载量:  63
  • 被引次数: 0
出版历程
  • 收稿日期:  2021-12-08
  • 修回日期:  2022-01-27
  • 上网日期:  2022-02-02
  • 刊出日期:  2022-05-20

/

返回文章
返回