基于最小二乘支持向量机的耕地面积预测研究

Cultivated land prediction based on least square support vector machines

针对耕地面积数据的小样本、复杂非线性特点,提出一种基于最小二乘支持向量机的耕地面积预测方法。采用相关系数法选择耕地面积的影响因子,通过粒子群优化算法对最小二乘支持向量机参数进行优化,最后建立耕地面积与影响因子之间复杂的非线性关系模型。采用湖南省耕地面积数据对模型性能进行验证,结果表明,相对于参比模型,最小二乘支持向量机提高了耕地面积的预测精度,是一种有效的耕地面积预测方法。

For cultivated land area has small sample data,complex nonlinear characteristic,this paper proposed a cultivated land area method based on least squares support vector machines.Firstly,it selected the influence factors of cultivated land area by correlation coefficient method,then optimized the parameters of least square support vector machines by particle swarm optimization algorithm,lastly built the complex nonlinear model between cultivated area and influence factors.It tested the proposed model by Hunan province cultivated land data,the results show that the proposed model improve the prediction accuracy of cultivated land area compare with other cultivated land area prediction method;the proposed model is an effective prediction method for cultivated land area.