Multistep-Ahead Independent Prediction of Nonlinear Time Series Based on Independent Model
YANG Zhenming1,YUE Jiguang1,WANG Xiaobao2,XIAO Yunshi1
(1. College of Electronic and Information Engineering, Tongji University, Shanghai 201804, China;2. Shanghai Shentong Rail Transit Research and Consultancy Co., Ltd., Shanghai 201103, China)
A multistep-ahead independent prediction approach of nonlinear time series was proposed. The step-by-step recurrent approach and the independent approach were compared, and the influence of accumulative error on the performance of multistep-ahead prediction was analyzed. The recurrent neural network (RNN) was used to realize the independent prediction approach, and the predictive model of urban rail transit is built, trained and tested by MATLAB. The predictive results showed that the error of independent prediction was smaller than that of step-by-step recurrent approach. The advantages and disadvantages of each approach were analyzed.
杨臻明1,岳继光1,王晓保2,萧蕴诗1. 基于独立模型的非线性时间序列多步超前预测[J]. 上海交通大学学报（自然版）, .
YANG Zhenming1,YUE Jiguang1,WANG Xiaobao2,XIAO Yunshi1. Multistep-Ahead Independent Prediction of Nonlinear Time Series Based on Independent Model. J. Shanghai Jiaotong Univ.(Sci.) , 2013, 47(10): 1626-1631.
［1］Badjate S L, Dudul S V. Multi step ahead prediction of North and South hemisphere sun spots chaotic time series using focused time lagged recurrent neural network model［J］. WSEAS Transactions on Information Science and Applications, 2009, 6(4): 684693.
Zhang D Q, Ning X X, Liu X N. Onling prediction of nonlinear time series using RBF neural networks［J］. Control Theory & Applications, 2009, 26(2): 151155.
［3］Ma Q L, Zheng Q L, Peng H, et al. Multistepprediction of chaotic time series based on coevolutionary recurrent neural network［J］. Chinese Physics B, 2008, 17(2): 536542.
Peng J Z, Wang Y N, Wang J. Robust Hinfinity trackingcontrol for robotic system based on recurrent fuzzyneuralnetworks［J］. Control Theory & Applications, 2009, 6(4): 684693.
Yang Z M, Yue J G, Wang X B, Li X L. Prediction of urban rail transit power consumption based on regression model［J］. Urban Mass Transit, 2010, 13(12):2225.
［6］Zemouri R, Gouriveau R, Zerhouni N. Defining and applying prediction performance metrics on a recurrent NARX time series model［J］. Neurocomputing, 2010, 73(1315): 25062521.
［7］Ludvig O, Nunes U. Novel Maximummargin Training Algorithms for Supervised Neural Networks［J］. IEEE Transactions on Neural Networks, 2010, 21(6): 972984.
［8］Cai X D, Zhang N. Time series prediction with recurrent neural networks trained by a hybrid PSOEA algorithm［J］. Neurocomputing, 2007, 70(1315): 23422353.
