BOD-DO Concentration Forecasting Model Based on Hybrid TS-SVM

Abstract

A hybrid TS-SVM model is provided, in which taboo search (TS) was used to optimize the key parameters of support vector machines (SVM) to make enhancement on the forecasting effect of support vector machines on BOD and DO. The hybrid TS-SVM model was used in forecasting the concentration of the indexes BOD and DO. Then by forecasting the concentration of the indexes BOD and DO in two cases. The results show that this method not only provides a simple and practical method for water quality forecasting, but also has a better forecasting effect, especially for small sample forecasting problem.