基于随机森林算法的海南岛水资源承载力自动评价方法  

Automatic Evaluation Method of Water Resources Carrying Capacity of Hainan Island Based on Random Forest Algorithm

作　　者：杨博雄 顾煜烨 李社蕾[1,2] 周波 汪舜敏 张开存 YANG Boxiong;GU Yuye;LI Shelei;ZHOU Bo;WANG Shunmin;ZHANG Kaicun(School of Information&Intelligence Engineering,University of Sanya,Sanya Hainan 572022,China;Academician Chen Guoliang Team Innovation Center,University of Sanya,Sanya Hainan 572022,China)

机构地区：[1]三亚学院信息与智能工程学院,海南三亚572022 [2]三亚学院三亚学院陈国良院士团队创新中心,海南三亚572022

出　　处：《海南热带海洋学院学报》2023年第5期82-90,共9页Journal of Hainan Tropical Ocean University

基　　金：海南省科技专项(ZDYF2022GXJS005)。

摘　　要：水资源是城市最重要的生态指标之一,尤其是与人类生活密切相关的淡水资源。首先分析影响城市水资源承载力(Water resources carrying capacity,WRCC)的淡水资源、社会、生态等主要因素,并细化WRCC关联因素的评价指标。采用基于分类回归树(Classification and regression tree,CART)的随机森林(Random forest,RF)机器学习方法,通过收集和整理历史数据,建立和训练WRCC智能评价模型,将训练好的模型应用于WRCC的自动评估。然后将海南岛18个市县的WRCC数据分成训练集和测试集,把训练集导入模型进行训练,并输出经训练的模型,再由测试集对其进行评估和检验。结果表明,基于CART的RF机器学习方法能够较好地拟合WRCC实际数据值与相应等级之间的变化规律;海南岛近年WRCC良(II)级的市县数量呈下降趋势,大部分市县WRCC呈现波动性特征。该自动评价方法可为城市管理者规划水资源可持续开发与环境保护提供决策依据,对其他资源开发与环境评估亦具有重要的借鉴意义。Water resource,especially the freshwater resource,which are closely related to human life,is one of the most important ecological indicators of a city.First analyzed were the main factors—fresh water resources,society,ecology,etc.—that affect water resources carrying capacity(WRCC)of a city,and the evaluation indicators of WRCC related factors were refined.The random forest(RF)machine learning method based on classification and regression tree(CART)was adopted to establish and train WRCC intelligent evaluation model by collecting and sorting historical data.And then the trained model was applied to the automatic evaluation of WRCC.WRCC data of 18 cities and counties in Hainan Island were divided into training set and test set.The training set was imported into the model for training,and the trained model was output which,later,was evaluated and tested by the test set.The results showed that the CART-based RF machine learning method can better fit the variation pattern between the actual data value of WRCC and the corresponding grades.The number of cities and counties with good(II)WRCC grade in Hainan Island has been decreasing in recent years,and most of the cities and counties showed fluctuating characteristics of WRCC.The current automatic evaluation method can provide a decision basis for city managers to plan sustainable water resources development and environmental protection,and also has important implications for other resource development and environmental assessment.

关 键 词：水资源承载力 机器学习 随机森林 分类回归 智能评价模型 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]