论文信息 - Simulation of Time Series Prediction Based on Smooth Support Vector Regression

Simulation of Time Series Prediction Based on Smooth Support Vector Regression

Time series analysis and prediction is an important means of dynamic system modelling, but traditional methods of time series prediction such as statistics and artificial neural network (ANN) are not fit for complicated non-linear system. Hence, a new method of support vector regression (SVR) was introduced to solve the prediction problem of complicated time series. For the purpose of reducing complexity of calculation, smooth arithmetic based on SVR was imported to forecast the time series of vibration data collected from turbine system. The result of simulation indicated that smooth support vector regression (SSVR) is obviously superior to ANN method on performance of prediction. Compared with SVR, SSVR has faster speed of convergence and higher fitting precision, which effectively extends the application of support vector machine. Keywords: time series prediction, support vector machine, regression, smooth method, turbine.

Chao Zhang | Guiji Tang | Pu Han | Guori Ji

[1] Xiao Jian-hua. The Application of SVR to Prediction of Chaotic Time Series , 2004 .

[2] Yang Bingru. Research on Complicated Time Series Prediction Based on Support Vector Machines , 2005 .

[3] Quan Yong. Successive Overrelaxation for Support Vector Regression , 2004 .

[4] Yuh-Jye Lee,et al. SSVM: A Smooth Support Vector Machine for Classification , 2001, Comput. Optim. Appl..

[5] David R. Musicant,et al. Successive overrelaxation for support vector machines , 1999, IEEE Trans. Neural Networks.

[6] Cai Jia. Smooth support vector regression , 2006 .

[7] John C. Platt,et al. Fast training of support vector machines using sequential minimal optimization, advances in kernel methods , 1999 .

[8] Wu Tie-jun,et al. Support Vector Machines for Regression , 2003 .

[9] Vladimir N. Vapnik,et al. The Nature of Statistical Learning Theory , 2000, Statistics for Engineering and Information Science.

[10] Ramazan Gençay,et al. Nonlinear modelling and prediction with feedforward and recurrent networks , 1997 .

[11] Federico Girosi,et al. An improved training algorithm for support vector machines , 1997, Neural Networks for Signal Processing VII. Proceedings of the 1997 IEEE Signal Processing Society Workshop.