利用复介电常数和BPAT神经网络进行环境监测的可行性研究

土壤中的污染物成分复杂，其含量与复介电常数之间具有很强的非线性关系。以土壤样品复介电常数的实部、虚部分别作为输入，以其含水率、体密度和所含6种已知离子的浓度分别作为输出，建立BP人工神经网络。把吉泰兰地区的土壤样品数据分为训练样本集和检验样本集，网络训练后，其学习效果显示模型的性能很好，检验样本的预测结果也与实测值较好吻合，说明利用复介电常数和BP人工神经网络进行环境监测是一种好的方法。

Application Feasibility of Complex Permittivity and BP Artificial Neural Networks for to Environmental Monitoring

The content of contamination is complex in the soil, and it has strong nonlinear relation with complex permittivity. Taking the real part and imaginary part of complex permittivity as inputs and the containing water rate, density and concentration of six known ions as outputs,it is found that the BP artificial neural network. Dividing the soil sample of Gitailand into training sample and testing sample, the effect of study shows that the model performed well. The forecast results of the test swatch inosculates the real results well. All of these shows that using complex permittivity and BP artificial neural network to observe and monitor environment is a good method.