机器学习预测甘肃省2025年—2030年污染物产量增长及匹配处理设施的应用研究  

Application of Machine Learning to Forecast Pollutant Production Growth and Matching Treatment Facilities in Gansu Province,2025—2030

作　　者：黄慧 高磊 汤超 

机构地区：[1]甘肃省生态设计环境科学研究院

出　　处：《节能与环保》2024年第8期19-27,共9页Energy Conservation ＆ Environment Protection

基　　金：《甘肃省污染物集中式污染治理设施匹配化建设研究报告》(GSWP-JS-201903)。

摘　　要：城市生活污水和垃圾的合理有效处理对实现城市的可持续发展起着至关重要的作用,而城市人口则直接影响着城市的可持续发展。本文利用《甘肃省发展年鉴(2009—2020)》数据,建立了基于支持向量回归的多输入机器学习(ML)人口预测模型。研究结果表明,在采用15个不同输入变量下,ML模型的决策系数在0.90~0.99之间。其中有4座城市产生的污水将在2030年超过现有处理能力,6座城市的垃圾填埋场将在2030年超过最大设计容量。这项研究为未来环境基础设施建设的战略决策提供了启示,有助于政府部门提前做出调控,以减轻环境压力。Urban domestic wastewater and rubbish treatment play a crucial role in achieving sustainable urban development,while urban population directly affects the sustainable development of cities.In this paper,a multi-input machine learning(ML)population prediction model based on support vector regression is developed using data from Gansu Development Yearbook(2009—2020).The results of the study show that the decision coefficients of the ML model range from 0.90 to 0.99 under the use of 15 different input variables,in which the wastewater generated in four cities will exceed the current treatment capacity in 2030,and the landfills in six cities will exceed the maximum design capacity in 2030.This study provides insights into strategic decision-making for future environmental infrastructure development,which can help government departments to make regulation in advance to reduce environmental pressure.

关 键 词：污染治理 设施匹配 机器学习 Scikit-Learn 相关性系数 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] X321[自动化与计算机技术—控制科学与工程]