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

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.