污水流量与天气数据融合的贝叶斯服务人口预测  

Bayesian Service Population Prediction Based on Sewage Flow and Weather Data Fusion

作　　者：蔡惠民 曹扬 陶政坪 谢真强 CAI Hui-min;CAO Yang;TAO Zheng-ping;XIE Zhen-qiang(CETC Big Data Research Institute Co.,Ltd.,Guizhou 550022,China;National Engineering Research Center of Big Data Application to Improvement of Governance Capacity,Guizhou 550022,China;School of Intelligence and Computing,Tianjin University,Tianjin 300354,China)

机构地区：[1]中电科大数据研究院有限公司,贵州贵阳550022 [2]提升政府治理能力大数据应用技术国家工程研究中心,贵州贵阳550022 [3]天津大学智能与计算学部,天津300354

出　　处：《计算机技术与发展》2024年第8期181-188,共8页Computer Technology and Development

基　　金：2021年海南省重大科技计划项目“海南自贸港社会管理服务信息系统和ODAE协同关键技术研究及应用示范”项目(ZDKJ2021051);国家自然科学基金项目(U19B2020)。

摘　　要：传统基于污水日均流量及人均用水量的人口预测模型缺乏对天气因素的考虑,存在人口数量测算偏大等问题。为了综合考虑天气因素对污水日均流量的影响,提出了一种基于污水监测数据与天气数据融合的贝叶斯服务人口预测模型。通过引入天气影响因子,同质化、异质化天气影响因子转化率,天气因素对污水日均流量的贡献量等,构建基于贝叶斯方法的污水日均流量生成模型。基于随机变分推理,获得生成模型参数的后验分布,进而实现各污水处理厂服务区域的服务人口预测模型。该模型能抵消区域天气因素的综合影响水平,能更合理地实现污水厂服务区域的人口数量预测。同时,通过统计分析对比了同质化、异质化天气影响因子转化率估计,天气因素对污水日均流量的影响等。该服务人口预测模型能进一步支撑城市人口的态势感知,对提升社会治理能力有重要意义。The traditional population prediction model based on the daily average sewage flow and per capita water consumption lacks consideration of weather factors,and there are problems such as overestimation of population size.In order to comprehensively consider the influence of weather factors on the daily average sewage flow,a Bayesian service population prediction model based on sewage monitoring data and weather data fusion is proposed.By introducing weather influence factors,the homogeneous and heterogeneous conversion rates of weather influence factors,and the contribution of weather factors to the daily average sewage flow,a Bayesian method based generative model is constructed.Based on stochastic variational inference,posterior distributions of the generative model parameters are obtained,and a service population prediction model for the service area of each sewage treatment plant is implemented.This model can offset the comprehensive influence level of regional weather factors and more reasonably predict the population of the sewage treatment plant service area.At the same time,statistical analysis was conducted to compare the estimation of homogeneous and heterogeneous conversion rates of weather impact factors,as well as the influence of weather factors on the daily average sewage flow rate.This service population prediction model can further support the perception of urban population trends and is of great significance for improving social governance capabilities.

关 键 词：污水监测 多源数据融合 服务人口预测 贝叶斯分析 随机变分推理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]