基于PSO-SVM模型的土壤适宜性评价  

Soil Suitability Evaluation Based on PSO-SVM Model

作　　者：王亚云[1] 赵艳玲[1] 何厅厅[1] 夏清[1] 侯占东[1] 石娟娟[1] 刘亚萍[1] 

机构地区：[1]中国矿业大学(北京)土地复垦与生态重建研究所,北京100083

出　　处：《河南农业科学》2013年第9期49-53,共5页Journal of Henan Agricultural Sciences

基　　金：国土资源部公益性行业专项(201111010-05)

摘　　要：土壤适宜性评价是获取土地质量状况的重要手段,可为土地科学规划、管理、科学决策提供重要依据。鉴于此,将支持向量机理论引入土壤评价领域,提出一个全新的土壤适宜性评价模型,为了提高评价精度,针对人为选择惩罚系数(C)、核函数参数(σ)的随机性,利用粒子群算法(PSO)对其进行优化,构建了PSO-SVM模型,SVM模型采用径向基函数(RBF)作为核函数。以溪洛渡水电站嘎勒移民安置区为例,利用PSO-SVM模型对土壤适宜性进行评价,同时与BP神经网络、普通SVM模型进行比较。结果表明:PSO-SVM算法明显提高了分类正确率,结果优于BP神经网络和普通SVM,能更好地反映土壤适宜性,可见,PSO-SVM是一种高精度的土壤适宜性评价模型。Soil suitability evaluation is an important means to obtain land quality status, and can provide important basis for the land planning, management and decision-making. In this paper, the support vector machine （SVM） theory is introduced to the field of soil suitability evaluation. Ai- ming at the randomness of artificial selection punishment coefficient （C） and kernel function pa- rameter（a）, a PSO-SVM model is constructed by using particle swarm optimization（PSO） to achieve higher evaluation accuracy. The radial basis function（RBF） is used as the kernel function in SVM model. Based on the model construction,Xiluodu Hydropower Station Gullah Resettlement Area was chosen as an example~ the PSO-SVM model was applied to evaluate soil suitability,and was compared with BP neural network and normal SVM model. The results showed that PSO- SVM significantly improved the accuracy of soil classification,compared with BP neural network and the SVM model. Therefore,PSO-SVM is a high-precision soil suitability evaluation model.

关 键 词：支持向量机 粒子群 综合评价 土壤适宜性 

分 类 号：S-03[农业科学]