基于小波神经网络的饮用水水质预测模型研究  

Research of predictive model of drinking water quality based on wavelet neural network

作　　者：邓宇[1] 姚宁[1] 田考聪[1] 

机构地区：[1]重庆医科大学公共卫生学院卫生统计教研室,重庆400016

出　　处：《重庆医科大学学报》2009年第4期455-458,共4页Journal of Chongqing Medical University

摘　　要：目的:为饮用水水质预测提供方法学参考。方法:采用小波神经网络(Wavelet neural network,WNN)方法建立水质预测模型,对重庆市某主城区饮用水水质参数高锰酸钾月平均浓度进行预测,并同BP神经网络的预测结果进行比较分析。结果:采用均方根误差(Root mean square error,RMSE)和平均绝对百分比误差(Mean absolute percentage error,MAPE)来评价模型预测效果。预测实例表明WNN模型具有较高的预测精度,并且预测精度高于BP神经网络。结论:利用WNN对水质参数具有较好的预测效果。Objective： To provide the methodology reference for the drinking water quality prediction. Methods： A predictive model of drinking water quality was established by wavelet neural network. The monthly average concentration of potassium permanganate in Chongqing, one drinking water quality parameter,was predicted by the model, and the predictive results were compared with BP neural network. Results： RMSE and MAPE were applied to evaluate the predictive results. The research indicated that the precision of WNN model was superior to that of BP neural network model. Conclusion： The WNN model has better precision for drinking water quality prediction.

关 键 词：小波神经网络 小波分析 BP神经网络 水质预测 

分 类 号：R123.1[医药卫生—环境卫生学]