基于改进差分进化和回声状态网络的时间序列预测研究

许美玲; 王依雯

doi:10.16383/j.aas.c180549

基于改进差分进化和回声状态网络的时间序列预测研究

doi: 10.16383/j.aas.c180549

许美玲^{1, 2,},
王依雯^1,

1.
大连理工大学电子信息与电气工程学部大连 116023
2.
东北大学智能工业数据解析与优化教育部重点实验室沈阳 110004

基金项目: 国家自然科学基金(61702077), 中央高校基本科研业务费(DUT16RC(3)123)资助

详细信息

作者简介:
许美玲：东北大学智能工业数据解析与优化教育部重点实验室讲师, 大连理工大学电子信息与电气工程学部博士后. 主要研究方向为神经网络和多元时间序列预测. 本文通信作者. E-mail: xuml@dlut.edu.cn

王依雯：大连理工大学电子信息与电气工程学部硕士研究生. 主要研究方向为时间序列预测. E-mail: wangyiwen0802@163.com

计量
- 文章访问数: 4950
- HTML全文浏览量: 3607
- PDF下载量: 324
- 被引次数: 0
出版历程
- 收稿日期: 2018-08-15
- 网络出版日期: 2019-12-28
- 刊出日期: 2021-07-27

Time Series Prediction Based on Improved Differential Evolution and Echo State Network

XU Mei-Ling^{1, 2
,},
WANG Yi-Wen^1
,

1.
Faculty of Electronic Information and Electrical Engineering, Dalian University of Technology, Dalian 116023
2.
Key Laboratory of Data Analytics and Optimization for Smart Industry (Northeastern University), Ministry of Education, Shenyang 110004

Funds: Supported by National Natural Science Foundation of China(61702077), Fundamental Research Funds for the Central Universities (DUT16RC(3)123)

More Information

Author Bio:
XU Mei-Ling　Lecturer at the Key Laboratory of Data Analytics and Optimization for Smart Industry (Northeastern University), Ministry of Education. Postdoctor at the Faculty of Electronic Information and Electrical Engineering, Dalian University of Technology. Her research interest covers neural networks and multivariate time series prediction. Corresponding author of this paper

WANG Yi-Wen　Master student at the Faculty of Electronic Information and Electrical Engineering, Dalian University of Technology. Her main research interest is time series prediction

摘要

摘要:
针对回声状态网络无法根据不同的时间序列有效地选择储备池参数的问题, 本文提出一种新型预测模型, 利用改进的差分进化算法来优化回声状态网络. 其中差分进化算法的缩放因子F、交叉概率CR和变异策略自适应调整, 以提高算法的寻优性能. 为验证本文方法的有效性, 对Lorenz时间序列、大连月平均气温 − 降雨量数据集进行仿真实验. 由实验结果可知, 本文提出的模型可以提高时间序列的预测精度, 且具有良好的泛化能力及实际应用价值.
- 时间序列 /
- 预测模型 /
- 差分进化 /
- 回声状态网络
Abstract:
For the echo state network, it is difficult to select the suitable reservoir parameters for different time series. In this paper, we propose a new prediction model which uses an improved differential evolution algorithm to optimize the parameters of the echo state network. The scale factor and crossover probability of differential evolution algorithms are adaptively adjusted. In addition, offspring generation strategy is also adaptively adjusted. To verify the effectiveness of the proposed method, experiments were conducted on Lorenz time series, and monthly average temperature - rainfall time series of Dalian. Experimental results show that the model can improve forecast accuracy and has good generalization ability and practicability.
- Time series /
- predictive model /
- differential evolution /
- echo state network

HTML全文

图 1 ESN结构示意图

Fig. 1 Structure of ESN

下载: 全尺寸图片幻灯片

图 2 IDE-ESN算法流程图

Fig. 2 Flow chart for IDE-ESN

下载: 全尺寸图片幻灯片

图 3 Lorenz-x(t)序列: IDE-ESN的预测曲线及误差曲线

Fig. 3 Lorenz-x(t) series: prediction and error curves obtained by IDE-ESN

下载: 全尺寸图片幻灯片

图 4 Lorenz-x(t)序列: 不同模型的适应度曲线

Fig. 4 Lorenz-x(t) series: the curves of fitness for different models

下载: 全尺寸图片幻灯片

图 5 大连月平均气温: IDE-ESN的预测曲线及误差曲线

Fig. 5 Dalian monthly average temperature series: prediction and error curves obtained by IDE-ESN

下载: 全尺寸图片幻灯片

图 6 大连月平均气温: 不同模型的适应度曲线

Fig. 6 Dalian monthly average temperature series: the curves of Fitness for differential models

下载: 全尺寸图片幻灯片

表 1 Lorenz-x(t)序列: IDE-ESN模型参数

Table 1 Lorenz-x(t) series: parameters in IDE-ESN

储备池参数	取值
储备池规模	50
稀疏度	0.0210
谱半径	0.9589
输入变化因子	0.0600

下载: 导出CSV

表 2 Lorenz-x(t) 序列: 测试集仿真结果

Table 2 Lorenz-x(t) time series: prediction results on the test dataset

模型	RMSE	SMAPE	NRMSE
AF-ESN	2.0850E-06	1.8571E-07	2.7992E-07
PSO-ESN	1.0139E-06	1.0211E-07	1.3613E-07
ELM	1.8422E-03	6.6638E-04	2.1061E-04
TLBO-ESN	7.7210E-07	1.6737E-07	1.0528E-07
IDE-ESN	3.2156E-07	9.8008E-08	4.3089E-08

下载: 导出CSV

表 3 Lorenz-x(t) 序列: 不同模型的运行时间

Table 3 Lorenz-x(t) series: run time of different models

模型	AF-ESN	PSO-ESN	TLBO-ESN	IDE-ESN
时间	1405.4289 s	47.6972 s	168.3124 s	102.8856 s

下载: 导出CSV

表 4 大连月平均气温: IDE-ESN模型参数

Table 4 Dalian monthly average temperature-rainfall series: parameters in IDE-ESN

储备池参数	取值
储备池规模	47
稀疏度	0.0206
谱半径	0.9802
输入变化因子	0.0459

下载: 导出CSV

表 5 大连月平均气温: 测试集仿真结果

Table 5 Dalian monthly average temperature series: prediction results for the test dataset

模型	RMSE	SMAPE	NRMSE
AF-ESN	1.8042	0.2902	0.1820
PSO-ESN	1.6511	0.2956	0.1666
ELM	5.4235	0.6704	0.5520
TLBO-ESN	1.6726	0.2088	0.1708
IDE-ESN	1.4215	0.2741	0.1440

下载: 导出CSV

表 6 大连月平均气温: 不同模型的运行时间

Table 6 Dalian monthly average temperature series: run time of different models

模型	AF-ESN	PSO-ESN	TLBO-ESN	IDE-ESN
时间	347.1955 s	10.5115 s	31.1971 s	15.1921 s

下载: 导出CSV

参考文献(30)

[1]	Su C H, Cheng C H. A hybrid fuzzy time series model based on ANFIS and integrated nonlinear feature selection method for forecasting stock. Neurocomputing, 2016, 205: 264−273 doi: 10.1016/j.neucom.2016.03.068
[2]	Haidar A, Verma B. A novel approach for optimizing climate features and network parameters in rainfall forecasting. Soft Computing, 2018, 22(24): 8119−8130 doi: 10.1007/s00500-017-2756-7
[3]	Arthun M, Eldevik T, Viste E, Drange H, Furevik T, Johnson H L, et al. Skillful prediction of northern climate provided by the ocean. Nature Communications, 2017, 8: 15875 doi: 10.1038/ncomms15875
[4]	Wang X, Jiang R, Li L, Lin Y, Zheng X, Wang F. Capturing car-following behaviors by deep learning. IEEE Transactions on Intelligent Transportation Systems, 2018, 19(3): 910−920 doi: 10.1109/TITS.2017.2706963
[5]	周平, 刘记平. 基于数据驱动多输出ARMAX建模的高炉十字测温中心温度在线估计. 自动化学报, 2018, 44(3): 552−561 Zhou Ping, Liu Ji-Ping. Data-driven multi-output ARMAX model for online estimation of central temperatures for cross temperature measuring in blast furnace ironmaking. Acta Automatica Sinica, 2018, 44(3): 552−561
[6]	张颜颜, 唐立新. 改进的数据驱动子空间算法求解钢铁企业能源预测问题. 控制理论与应用, 2012, 29(12): 1616−1622 Zhang Yan-Yan, Tang Li-Xin. Improved data-driven subspace algorithm for energy prediction in iron and steel industry. Control Theory and Application, 2012, 29(12): 1616−1622
[7]	郑念祖, 丁进良. 基于Regression GAN的原油总氢物性预测方法. 自动化学报, 2018, 44(5): 915−921 Zheng Nian-Zu, Ding Jin-Liang. Regression GAN based prediction for physical of total hydrogen in crude oil. Acta Automatica Sinica, 2018, 44(5): 915−921
[8]	Hearst M A, Dumais S T, Osuna E, Platt J, Scholkopf B. Support vector machines. IEEE Intelligent Systems and Their Applications, 1998, 13(4): 18−28 doi: 10.1109/5254.708428
[9]	Trappey A J, Hsu F C, Trappey C V, Lin C I. Development of a patent document classification and search platform using a back-propagation network. Expert Systems with Applications, 2006, 31(4): 755−765 doi: 10.1016/j.eswa.2006.01.013
[10]	Huang G B, Zhu Q Y, Siew C K. Extreme learning machine: Theory and application. Neurocomputing, 2006, 70: 489−502 doi: 10.1016/j.neucom.2005.12.126
[11]	Jaeger H, Haas H. Harnessing nonlinearity: Predicting chaotic systems and saving energy in wireless communication. Science, 2004, 304(5667): 78−80 doi: 10.1126/science.1091277
[12]	韩敏, 任伟杰, 许美玲. 一种基于L1范数正则化的回声状态网络. 自动化学报, 2014, 40(11): 2428−2435 Han Min, Ren Wei-Jie, Xu Mei-Ling. An improved echo state network via L1-norm regularization. Acta Automatica Sinica, 2014, 40(11): 2428−2435
[13]	伦淑娴, 林健, 姚显双. 基于小世界回声状态网的时间序列预测. 自动化学报, 2015, 41(9): 1669−1679 Lun Shu-Xian, Lin Jian, Yao Xian-Shuang. Time series prediction with an improved echo state network using small world network. Acta Automatica Sinica, 2015, 41(9): 1669−1679
[14]	Han M, Xu M. Laplacian echo state network for multivariate time series prediction. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29(1): 238−244 doi: 10.1109/TNNLS.2016.2574963
[15]	李德才, 韩敏. 基于鲁棒回声状态网络的混沌时间序列预测研究. 物理学报, 2011, 60(10): 108903 doi: 10.7498/aps.60.108903 Li De-Cai, Han Min. Chaotic time series prediction based on robust echo state network. Acta Physica Sinica, 2011, 60(10): 108903 doi: 10.7498/aps.60.108903
[16]	韩敏, 王亚楠. 基于Kalman滤波的储备池多元时间序列在线预报器. 自动化学报, 2010, 36(1): 169−173 doi: 10.3724/SP.J.1004.2010.00169 Han Min, Wang Ya-Nan. Multivariate time series online predictor with Kalman filter trained reservoir. Acta Automatica Sinica, 2010, 36(1): 169−173 doi: 10.3724/SP.J.1004.2010.00169
[17]	Xu D, Lan J, Principe J C. Direct adaptive control: an echo state network and genetic algorithm approach[C]. In: Proceedings of the 2005 IEEE International Joint Conference on Neural Networks. Montreal, Canada: IEEE, 2005. 3: 1483−1486
[18]	Wang J S, Han S, Guo Q P. Echo state networks based predictive model of vinyl chloride monomer convention velocity optimized by artificial fish swarm algorithm. Soft Computing, 2014, 18(3): 457−468 doi: 10.1007/s00500-013-1068-9
[19]	Chouikhi N, Ammar B, Rokbani N, Alimi A M. PSO-based analysis of echo state network parameters for time series forecasting. Applied Soft Computing, 2017, 55: 211−225 doi: 10.1016/j.asoc.2017.01.049
[20]	孙晓燕, 陈姗姗, 巩敦卫, 张勇. 基于区间适应值交互式遗传算法的加权多输出高斯过程代理模型. 自动化学报, 2014, 40(2): 172−184 Sun Xiao-Yan, Chen Shan-Shan, Gong Dun-Wei, Zhang Yong. Weighted multi-output Gaussian process-based surrogate of interactive genetic algorithm with individual′s interval fitness. Acta Automatica Sinica, 2014, 40(2): 172−184
[21]	Du W, Ying W, Yan G, Zhu Y, Cao X. Heterogeneous strategy particle swarm optimization. IEEE Transactions on Circuits and Systems II: Express Briefs, 2017, 64(4): 467−471 doi: 10.1109/TCSII.2016.2595597
[22]	Mavrovouniotis M, Muller F M, Yang S. Ant colony optimization with local search for dynamic traveling salesman problems. IEEE Transactions on Cybernetics, 2017, 47(7): 1743−1756 doi: 10.1109/TCYB.2016.2556742
[23]	Zhang Z, Wang K, Zhu L, Wang Y. A Pareto improved artificial fish swarm algorithm for solving a multi-objective fuzzy disassembly line balancing problem. Expert Systems with Applications, 2017, 86: 165−176 doi: 10.1016/j.eswa.2017.05.053
[24]	胡蓉, 钱斌. 一种求解随机有限缓冲区流水线调度的混合差分进化算法. 自动化学报, 2009, 35(12): 1580−1586 Hu Rong, Qian Bin. A hybrid differential evolution algorithm for stochastic flow shop scheduling with limited buffers. Acta Automatica Sinica, 2009, 35(12): 1580−1586
[25]	Rao R V, Savsani V J, Vakharia D P. Teaching-learning-based optimization: A novel method for constrained mechanical design optimization problems. Computer-Aided Design, 2011, 43: 303−315 doi: 10.1016/j.cad.2010.12.015
[26]	Xu M L, Han M, Lin H F. Wavelet-denoising multiple echo state networks for multivariate time series prediction. Information Sciences, 2018, 465: 439−458 doi: 10.1016/j.ins.2018.07.015
[27]	周晓根, 张贵军, 郝小虎. 局部抽象凸区域剖分差分进化算法. 自动化学报, 2015, 41(7): 1315−1327 Zhou Xiao-Gen, Zhang Gui-Jun, Hao Xiao-Hu. Differential evolution algorithm with local abstract convex region partition. Acta Automatica Sinica, 2015, 41(7): 1315−1327
[28]	丁进良, 杨翠娥, 陈立鹏, 柴天佑. 基于参考点预测的动态多目标优化算法. 自动化学报, 2017, 43(2): 313−320 Ding Jin-Liang, Yang Cui-E, Chen Li-Peng, Chai Tian-You. Dynamic multi-objective optimization algorithm based on reference point prediction. Acta Automatica Sinica, 2017, 43(2): 313−320
[29]	Zhang J Q, Sanderson A C. JADE: Adaptive differential evolution with optional external archive. IEEE Transactions on Evolutionary Computation, 2009, 13(5): 945−958 doi: 10.1109/TEVC.2009.2014613
[30]	Zhao S Z, Suganthan P N, Das S. Self-adaptive differential evolution with multi-trajectory search for large-scale optimization. Soft Computing, 2011, 15: 2175−2185 doi: 10.1007/s00500-010-0645-4