摘要
为建立农用地(耕地)质量评价模型,客观准确地进行农用地(耕地)分等,减少现行农用地分等方法中的人为因素影响,提高农用地分等的精度。以福建省长泰县丘陵山地区为实证研究区,通过无监督网络——自组织特征映射网络(SOM)筛选出2 602组典型样本,分别进行有监督网络——BP神经网络和支持向量机(SVM)的学习训练,将分等指标作为输入变量,以农用地自然质量等指数和等别作为输出变量,分别建立BP神经网络农用地分等模型与SVM农用地分等模型并对其精度进行分析。BP神经网络模型的评价正确率为89%,精度较高;支持向量机(SVM)模型的评价结果正确率为99%,达到高精度等级。2种模型均能满足农用地分等的精度要求,但SVM模型较BP神经网络效果更好,更适合应用于农用地分等工作。
Improving the accuracy of agricultural land classification,through establish the quality evaluation model of agricultural land(cultivated land) can classify the agricultural land(cultivated land)objectively and accurately and reduce the impact of human factors in the current agricultural classification method.The hilly area of Changtai County in Fujian Province is used as an empirical area.MATLAB software is adopted as the operating platform.Supervised network learning(BP-ANN and SVM)and unsupervised network learning(SOM)methods are combined.Firstly,a total of 2 602 group typical samples are selected by using self-organizing feature mapping network(SOM)through unsupervised network learning.Network learning about BP neural network and support vector machine(SVM)are monitored.Seven indicators(effective soil thickness,organic matter content,slope,field road accessibility,altitude,soil texture and irrigation guarantee rate)are then used as input variables.Natural grade indexes or grade are used as output variables,the evaluation model of cultivated land quality of BP neural network and support vector machine(SVM)are established.The results show that:The evaluation accuracy of BP neural network method is high.The correct rate of total evaluation result is 89%.As for the SVM method,the correct rate of total evaluation result has remarkable accuracy of 99%.Both methods can meet the accuracy requirements of agricultural land classification.In the whole,SVM method is more effective than the BP neural network.It is more suitable for agricultural land classification.
作者
范胜龙
邱凌婧
茹凯丽
陈巧燕
胡勇
FAN Shenglong;QIU Lingjing;RU Kaili;CHEN Qiaoyan;HU Yong(College of Resource and Environmental Science,Fujian Agriculture and Forestry University,Fuzhou 350002,China;Geosctences Documentation Center of Ghana Geological Survey,Beijing 10083,China)
出处
《中国农业大学学报》
CAS
CSCD
北大核心
2018年第12期138-148,共11页
Journal of China Agricultural University
基金
福建省自然科学基金资助项目(2015J01624)
关键词
农用地分等
BP人工神经网络
支持向量机模型
agricultural land classification
back-propagation neural network
support vector machine