城镇智慧水务日用水量预测方法改进分析  

Analysis on Improvement of Daily Water Consumption Prediction Method of Urban Smart Water

作　　者：孙小燕 SUN Xiao-yan(Huizhou City Water Supply Co.Ltd.,Huizhou 516001,Guangdong,China)

机构地区：[1]惠州市供水有限公司,广东惠州516001

出　　处：《水利科技与经济》2022年第11期130-135,共6页Water Conservancy Science and Technology and Economy

摘　　要：随着5G技术的发展,智慧水务成为建设智慧城市的重要一环。因此,建立一种能够精确预测城市用水量的预测算法尤为关键。基于5G智慧水务技术,以H市自来水公司2020年的数据为研究对象,在考虑温度、节假日、天气情况等影响因素的条件下,对LR、SVR、BPNN这3种算法进行评估并改进。结果表明,通过改进SVR、BPNN算法的主要参数,改进后算法的预测精度明显较改进前有进一步提高;LR、BPNN算法受测试集与温度影响较小,但SVR算法受温度影响较大,且高温或低温均会减小其预测精度,但通过增加测试集可以降低此类情况。建议在一般情况下,采用改进后LR、BPNN算法;而在测试比例较大时,可以采用SVR算法进行用水量的预测。With the development of 5G technology,smart water affairs have become an important part of building a smart city,so it is particularly critical to establish a prediction algorithm that can accurately predict urban water consumption.Based on 5G smart water technology,this paper takes the data of a water company in City A in 2020 as the research object,and evaluates and improves the three algorithms of LR,SVR and BPNN under the conditions of temperature,holidays,weather conditions and other influencing factors.The error of the improved algorithm is compared and analyzed,and it is found that by improving the main parameters of the SVR and BPNN algorithms,the prediction accuracy of the improved algorithm is obviously further improved than that before the improvement;the LR and BPNN algorithms are less affected by the test set and temperature,However,the SVR algorithm is greatly affected by temperature,and high or low temperature will reduce its prediction accuracy.This situation can be reduced by increasing the test set.It is recommended to use the improved LR and BPNN algorithms in general,and can be used when the test ratio is large.The SVR algorithm is used to predict the water consumption.

关 键 词：智慧水务 预测算法 5G 用水量 

分 类 号：TV213[水利工程—水文学及水资源]