基于自组织特征映射神经网络的土壤分类

Soil Classification Based on Self-Organizing Feature Mapping Neural Networks

人工神经网络中的自组织特征映射网络具有较强的聚类功能,将自组织特征映射神经网络模型应用于土壤分类,提取影响土壤分类的七个理化因子,根据19个土壤样本建立神经网络,最后验证10个土壤样本的分类结果是否正确。分析结果表明,这种方法是十分有效和方便的。同时,本文对分类结果进行分析和讨论,指出利用该模型强大的学习功能及很好的自适应性、自组织性和鲁棒性可以为土壤分类提供一种快速、准确的信息处理手段。

The self-organizing feature mapping neural networks in ANN possess a strong clustering function. We apply the self-organizing feature mapping neural networks to soil classification by selecting seven factors related to soil classifition and building an ANN according to 19 samples. Finally we test the classification results of 10 samples. The test results show the way is very effective and convenient. In the meantime, this paper analyses the results of classification, and points out that SOFM can provide a rapid and accurate means for studying the soil classification system depending on its strong learning function, excellent self-organizing feature, self-adapting feature and robustness.