基于概率论和BP网络的水质评价与灰色动态预测模型

讨论了BP网络模型存在的不足。提出了建立合理的BP网络模型的基本原则和步骤。针对水质评价与预测问题。通过在各类水质污染指标浓度区间内生成随机分布样本的方法。组成足够多用于BP网络训练、检验和泛化能力评定用的样本。建立了南京市秦淮外河水体水质评价的BP网络模型：给出了区分不同类别水质的模型分界值样本和模型输出分界值。用概率分布理论分析处于相邻类别水质过渡状态水体的水质属性。建立的水质评价模型应用于实例表明。南京市秦淮外河水体1991—1996年水质为V类向IV类转变。1992年水质最差，1996年水质最好。灰色动态预测模型计算表明。从1997到2001年。南京市秦淮外河水体水质逐年变好。

On Water Quality Assessment and Prediction Model Applying BP Networks, Probability Theory and Gray Dynamic Theory

Discussion on the shortcomings of BP networks, and the essential principles and modeling procedure of establishing reasonable BP network have been put forward in this paper. As to water quality assessment and prediction, a new approach producing training set data, testing set data and critical values set data randomly distributed between the critical values were established. The water quality assessment model for outside sector of Qinghui River, in Nangjing City, was set up. The quality character of water lying to transition-state between two-connected Grade was analyzed by applying Normal Distribution and Probability. The water quality prediction model was established by applying Gray Dynamic theory. The water quality of 1991 ~ 1996 was changing from Grade V to IV, and it was the worst in 1992 and the best in 1996. The water quality was getting better and better from 1997 to 2001.