基于最小二乘支持向量机的人民胜利渠灌区灌溉用水量预测

基于最小二乘支持向量机（LSSVM）良好的泛化能力和特点，以人民胜利渠灌区需水量为研究对象，选用径向基函数（RBF）作为核函数，建立了最小二乘支持向量机预测模型，对灌区需水量进行了模拟计算，用检验样本与灰色预测和基于RBF的神经网络模型的预测结果进行了比较，LSSVM预测的最大误差8.78%，平均误差4.90%。结果表明最小二乘支持向量机模型有较高的预测精度和较强的泛化能力，可为灌区水资源规划提供科学依据。

Irrigation Water Prediction for People's Victory Canal Irrigation District based on Least Squares Support Vector Machine Model

Based on the good generalization ability of Support Vector Machine model, water consumption for the People's Victory Canal Irrigation district was studied. Radial basis function (RBF) was chosen as the kernel function, a least squares support vector machine (LSSVM) prediction model was set up. The irrigation water demands of the People’s Victory Canal Irrigation district for 8 years were simulated. The test samples were compared with the results of gray prediction and RBF-based neural network model. The maximum predicted error of SVM was 8.78%, and the average error of 4.90%. The results show that support vector machine model has higher prediction accuracy and stronger generalization ability. It can provide a scientific basis for irrigation water resources planning.