Entity

Time filter

Source Type


Wang X.,Southwest University | Wang X.,Key Laboratory of Chongqing Digital Agriculture | Lv J.,Southwest University | Lv J.,Key Laboratory of Chongqing Digital Agriculture | And 3 more authors.
ICIC Express Letters, Part B: Applications | Year: 2014

Soil moisture content prediction is very important for water management and planning. However, operation of soil moisture prediction is challenging due to high fluctuations, intermittent and stochastic nature of soil moisture series. In this paper, we develop a hybrid method to predict soil moisture content. The basic idea of the hybrid model is to eliminate the random fluctuation of soil moisture series by wavelet transform technology (WTT), and then the support vector machine (SVM) model is established to predict the approximation of soil moisture signal obtained by WTT. The parameters in SVM are fine tuned by genetic algorithm (GA) to ensure the generalization of SVM. Real data from one citrus orchard soil moisture series of Chongqing are used to evaluate the prediction accuracy. The results show that the hybrid method can improve the prediction performance in an effective way. © 2014 ICIC International. Source

Discover hidden collaborations