基于BP神经网络模型在珊溪水库水华预测中的应用  

The Application of BP Neural Network in Forecasting Algal Blooms of Shanxi Reservoir

作　　者：王珅 万哲慧 冯孙林 黄晶 刘秀峰 赵小敏 

机构地区：[1]浙江省温州市环境监测中心站,浙江温州325000 [2]山东省枣庄市环境监测中心站,山东枣庄277100

出　　处：《四川环境》2018年第1期39-43,共5页Sichuan Environment

基　　金：浙江省自然科学基金(LZ12C03001);温州市"水体污染控制与治理"科技创新项目(S20140024)资助

摘　　要：在珊溪水库藻类暴发期间应急监测数据的基础上,建立pH值、高锰酸盐指数、总氮、总磷、叶绿素a数据矩阵。运用MATLAB R2015b GUI可视化界面模块,将应急监测数据样本空间分为训练样本、验证样本、测试样本,建立珊溪水库BP神经网络模型,预测了珊溪水库藻类暴发期间叶绿素a浓度。BP神经网络建模结果显示:输出数据与实测数据相关系数0.978,平均相对误差-0.19%,标准方差18.54%,模型稳定性较好,叶绿素a预测结果符合预期。BP神经网络预测模型为珊溪水库饮用水水源地环境保护提供了科学依据。Based on the emergency monitoring data of algal blooms period in Shanxi reservoir, the data matrix of pH, C0DMn, TN, TP and Chi. a was established. Using MATLAB R2015b GUI visual interface module, the emergency monitoring data matrix was divided into training samples, validation samples and test samples in order to establish the BP neural network model and predict the concentration of Chi. a during the period of algae blooms. The results showed that the correlation coefficient between the output data and the measured data was 0. 978 , the average relative error was - 0. 19% , the standard deviation was 18. 54%. The model has good stability, the prediction of Chi. a meet expectation. BP neural network prediction model provided a scientific basis for the protection of Shanxi reservoir.

关 键 词：MATLAB BP神经网络 珊溪水库 水华预测 

分 类 号：X524[环境科学与工程—环境工程]