农业灌溉用水量的LS-SVM预测模型研究

Forecasting Irrigation Water Requirement Based on Least Squares Support Vector Machine

农业灌溉用水量预报是灌区制定水资源调度计划、合理高效分配水量的科学依据。针对灌溉用水量影响因素复杂非线性的特点,鉴于支持向量机算法的诸多优势,建立了基于最小二乘支持向量机的灌溉用水量预测模型,将该模型用于塔河流域T灌区灌溉用水量预测,并与人工神经网络方法预报结果比较,表明该方法具有泛化能力强、误差小等特点。

The irrigation water requirement forecast is the basis for making scheduling program of water resource and allocating on water in irrigation area rationally and efficiently.The factors influencing the irrigation water are complex and nonlinear,and SVM has many advantages on nonlinear small samples,this paper introduced SVM into forecasting irrigation water requirement and proposed a forecasting model of irrigation water requirement based on LS-SVM.In case study,the forecasting model was applied to estimate the irrigation water requirement of T irrigation area in Tarim River Basin,and was compared with BP artificial neural network（BP-ANN）.The result indicated that the forecast model based on LS-SVM has an excellent generalization ability and small error.LS-SVM provides an effective method to forecast irrigation water requirement.