基于案例推理和机器学习的场地污染风险管控与修复方案推荐系统构建技术  

Construction technology for site pollution risk control and remediation scheme recommendation system supported by case-based reasoning and machine learning

作　　者：张秋垒 黄国鑫 王夏晖 毕二平[2] 季国华 陈茜 卢然 ZHANG Qiulei;HUANG Guoxin;WANG Xiahui;BI Erping;JI Guohua;CHEN Xi;LU Ran(Chinese Academy of Environmental Planning;School of Water Resources and Environment,China University of Geosciences(Beijing))

机构地区：[1]生态环境部环境规划院 [2]中国地质大学(北京)水资源与环境学院

出　　处：《环境工程技术学报》2020年第6期1012-1021,共10页Journal of Environmental Engineering Technology

基　　金：国家重点研发计划项目(2018YFC1800205)。

摘　　要：为解决我国现有场地土壤和地下水污染风险管控与修复方案选择不合理、筛选效率低等问题,通过结构化层次存储和搜索技术,运用案例推理和机器学习,构建场地污染风险管控与修复方案推荐系统,实现快速搜索查找匹配源案例;通过研究案例库实现途径和内容,进行方案推荐系统的结构设计和系统开发,建立了基于万维网技术的案例检索查询页面;采用K最近邻算法和层次分析法在案例库中检索相似案例,实现了方案推荐功能,以期为决策者选取相对优化的场地污染治理方案提供参考,提高我国场地污染风险管理的精准化、智能化、高效化和低成本化。In order to solve the problems of unreasonable selection and low screening efficiency of the existing soil and groundwater pollution risk control and remediation schemes in China,the site pollution risk control and remediation scheme recommendation system was constructed by using the structured hierarchical storage and search technology as well as case-based reasoning(CBR)and machine learning methods,which realized fast search to find matching source cases.By studying the realization way and content of case base,the structure design and system development of the scheme recommendation system were carried out,and the case retrieval query page based on Web technology was established.Similar cases were retrieved in the case database by using K-nearest neighbor algorithm and analytic hierarchy process,and the scheme recommendation function was realized,in order to provide reference for decision-makers to select relatively optimized site pollution control scheme,and improve the accuracy,intelligence,high efficiency and low cost of site pollution risk management in China.

关 键 词：机器学习 场地污染 案例推理 风险管控 方案推荐 

分 类 号：X32[环境科学与工程—环境工程]