基于MATLAB的BP网络在月生活用水量预测中的应用  

Application of the prediction of monthly domestic water consumption based on MATLAB BP neural network

作　　者：吴限[1] 吴恬[1] 隋顺彬[2] 刘光民[3] 雍子豪[1] 

机构地区：[1]中国人民解放军后勤工程学院,重庆401311 [2]中国人民解放军第四军医大学,陕西西安714200 [3]武警总医院,北京100039

出　　处：《供水技术》2011年第6期32-36,共5页Water Technology

摘　　要：月生活用水量的预测对于供水企业制定月用水量生产计划具有重要的指导意义。以重庆市某建筑的实际用水量为样本,建立了传递函数为logsig、训练函数为trainlm的3-10-1结构的BP网络模型,计算出网格的训练误差为0.044 4,训练样本的平均相对误差的绝对值为1.35%,并应用该模型预测了重庆市某小区建筑2011年3—6月的月生活用水量。It is of great guiding significance for water supply companies to make monthly production plan based on the prediction of monthly domestic water consumption. Taking the actual water consumption of a building in Chongqing City as an example, the structure of 3 - 10 - 1 ＇ s BP network model was established based on the logsig and trainlm. The training error of the grid was 0. 044 4. The absolute value of the mean relative error （MRE） of trained sample was 1.35%. Finally the monthly domestic water consumption of a building in Chongqing City from March to June in 2011 was predicted based on the model.

关 键 词：MATLAB BP网络 月生活用水量 用水量预测 

分 类 号：TU991.31[建筑科学—市政工程]