BP神经网络模型在太湖出入湖河流水质预测中的应用

摘要

利用BP（Back Propagation）神经网络模型预测太湖出入湖河道水质污染指数，为太湖流域管理规划平台提供决策支持。建立三层BP神经网络预测模型，以1996年到2004年太湖例行环境监测数据为基础，进行水质预测，研究出入湖河道与太湖水质变化的关系。对太湖2005年的水质情况进行预测，结果表明，2005年水质污染情况有所改善，基本为V类水质，符合太湖水质污染情况发展态势。传统的数学建模方法要首先确定待定参数，而在实际应用中，许多不可测量估计的因素会对结果造成很大的影响。BP神经网络适应性较好、预测精度较高，能较好地反应水质指标的内在变化规律，为控制水环境污染提供科学依据。%In order to provide decision-making support to management and planning platform of Taihu Lake basin , we use BP ( back propagation) neural network model to predict the water contamination index of rivers inflowing to and outflowing from Taihu Lake .We establish a three-layer BP neural network prediction model to predict water quality on the basis of routine environmental monitoring data of Taihu from year 1996 to year 2004, and study the relationship between inflowing and outflowing rivers and the variation of water quality of Taihu.The water quality of Taihu in year 2005 was predicted, the result showed that in 2005 the water quality pollution had some amelioration than before , generally in grade V water quality , and was in line with the development trend of Taihu water pollution situation . Traditional mathematical modelling method requires making sure the unknown parameters first , but in practical application , multiple unmeasurable and unpredictable factors will greatly affect the result .BP neural network method has good adaptability and higher precision , it can better reflect the inherent variation rules of water quality indexes , and provides scientific basis for controlling the water environmental pollution.