Research on Hydrology Time Series Prediction Based on Grey Theory and [epsilon]-Support Vector Regression

Abstract

Hydrology time series prediction is significant. It is not only helpful to set the planning in daily configuration works of water resources, but also provides guidance for leaders to make decision, especially in some special case such as flood and seriously lack water. In order to solve the imbalance complexity of prediction model and complexity of samples and raise forecasting accuracy, combined prediction model based on support vector machine and grey theory was proposed. The grey time series prediction method was used to reduce complexity of samples and the support vector machine regression was used to reduce complexity of prediction model. The incoming water time series of Minjiang River in 1937-2002 were taken as the sample to be analyzed. The results show that the combined algorithm of ¦Å- support vector regression and grey theory has better effects in simulate of trend data and the random data in medium and long-term forecasting.