几种神经网络模型在空气质量评价中的应用

人工神经网络是一种利用计算机模拟人脑神经组织的运算模型。作为一种新的研究方法。神经网络在一定程度上可以弥补传统评价方法需要构建隶属函数、无法精确描述级别区间内的变化特征以及设计过程具有一定人为偏好的不足。本文选用BP网络、径向基函数网络、LVQ网络和Elman网络这4种典型的神经网络模型进行实例研究。把国内8个城市的污染物排放数据代入训练好的网络模型，进行空气质量评价，得出的结论对提高神经网络评价结果的准确性和可靠性有一定参考作用。

Application of Several ANN Models to Air Quality Assessment

Most of traditional methods applied to environmental quality assessment need to construct subject functions and confirm the weights of evaluation index.It will bring certain subjective effects to assessment processes and some important information will be lost.Artificial neural network(ANN) as a model simulating biological neural network of human brain by computer can remedy all these shortcomings.In this paper,we chose four classic networks to compare by inputting into trained networks the values of air pollution concentration of 8 cities in China,useful conclusions are drawn.