基于ANN的土壤重金属分布和污染评价研究

农田土壤重金属污染与备受关注的农产品安全问题有密切联系，因此对其进行研究意义重大。以江苏省南通市为研究区，利用采样点实测数据，借助神经网络模型(ANN)并结合3S技术对问题进行研究，从而对土壤重金属的空间动态分布进行描述，并对各个空间位点重金属的污染状况进行评价。结果表明，神经网络模型能够智能地学习各个样点的空间位置与该点各重金属含量之间的映射关系和预先设计好的分类评价模式，并能够稳健地对各个空间插值点处的重金属含量和各个位点的重金属污染状况进行预测和评价。结论显示，南通市大部分农田土壤重金属污染较轻，但也存在局部地区的严重污染。结论与实际情况相符，表明神经网络模型可以为农田土壤重金属的研究提供一个新的思路和方法。

DISTRIBUTION AND EVALUATION OF SOIL HEAVY METALS BASED ON ARTIFICIAL NEURAL NETWORKS

It is significant to study the pollution of heavy metals in soil,because the pollution is related with the safety of primary products.The objective of this paper is to describe the spatial dynamics of heavy metals in soils,and evaluate the pollution level of heavy metals in spatial scale.Thus,Nantong city in Jiangsu province was selected as the research region,and data obtained were analyzed using Artificial Neural Networks(ANN) Modeling and 3S technology.The results showed that the ANN modeling cannot intelligently reveal the relationship between spatial position and heavy metal content,and predict robustly the heavy metal contents in interpolating dots and reveal correctly the pollution level in every dot according to teacher signals.It is concluded that the farmland soil in large region in Nantong city should be grouped into lower pollution lever,but some fractions belong to severe pollution lever.They are coincident with the actual facts,so ANN modeling can provide a new thought and method to investigate the problem of soil heavy metals.