Simulation of time series prediction based on smooth support vector regression

Abstract

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. © Springer-Verlag Berlin Heidelberg 2007.