农业区地下水位动态变化预测的支

当观测资料的数据量少而又存在多个相互影响或关联的变量时，常用的回归预测模型不能全面考虑多个变量。在地下水位动态变化预测中应用了一种新的方法——支持向量机方法(SVM ) , 该方法属于机器学习理论发展的最新阶段, 具有专门针对有限样本、算法复杂度与样本维数无关等优点。针对一些农区井灌水稻规模扩大而引起地下水资源紧缺的情况，以某井灌水稻地区地下水动态观测资料为研究对象，运用支持向量回归模型，描述其地下水动态变化趋势。

A Support Vector Machine Model for the Groundwater Forecast in Agricultural Areas

The conventional regression model cannot take into account multi-variables overall for forecast problems which have several variables relating with each other with inadequate observational data available.In this paper,a novel modeling and estimation model of groundwater forecast is presented based on Support Vector Machine(SVM).The structure of SVM has many computation advantages,such as feasibility for a limited sample and irrelevance between the complexity of algorithm and the sample dimension.By applying SVM model with observed data of some agricultural districts,which are lack of groundwater because the scale of rice irrigated by groundwater is enlarged.Compared with the methods currently in use,the SVM method has a great potential superiority in predicting accuracy.