智能算法在生态学研究多元场景中的应用进展  

Progress in the application of intelligent algorithms in multiple ecological scenarios

作　　者：戈晓宇[1] 翟哲然 黄子玲 解圆圆 王海燕[2] 兰雨萌 王帅清 汶宣彤 GE Xiaoyu;ZHAI Zheran;HUANG Ziling;XIE Yuanyuan;WANG Haiyan;LAN Yumeng;WANG Shuaiqing;WEN Xuantong(School of Landscape Architecture,Beijing Forestry University,Beijing 100083,China;School of Information Science and Technology,Beijing Forestry University,Beijing 100083,China)

机构地区：[1]北京林业大学园林学院,北京100083 [2]北京林业大学信息学院,北京100083

出　　处：《生态学报》2025年第2期1013-1047,共35页Acta Ecologica Sinica

基　　金：国家重点研发计划(2023YFF1304600);北京市共建项目(2015BLUREE01);北京市社会科学基金项目(21JCC094);国家自然科学基金项目(31800606)。

摘　　要：生态学研究领域中对智能算法的使用呈现越来越丰富的趋势,其解决了许多重要问题。智能算法的应用已逐渐成为生态学研究的重要话题。研究以中国知网(CNKI核心)和Web of Science核心数据库中42439篇智能算法在生态学领域应用的相关学术论文为依据,借助文献计量学软件CiteSpace.6.3R1,介绍2013—2023年间国内外研究热点的发展现状和情况;根据每种智能算法在生态学优化、预测和评估研究中的作用,分类论述其实际研究过程和应用特征;分析智能算法应用的优势和当前存在的局限性;回顾智能算法对生态学研究的意义,并提出了对未来发展前景的展望。The deployment of intelligent algorithms in ecological research is increasingly prevalent,addressing numerous significant issues.The application of intelligent algorithms is emerging as a critical area of focus within ecological research.Firstly,based on the CiteSpace.6.3R1 econometric analysis visualization software,42439 papers related to the application of intelligent algorithms in ecology from 2013 to 2023 and found in the CNKI(China National Knowledge Infrastructure)and WOS(Web of Science)core collections were taken as study objects to assess global and domestic trends and hotspots.Subsequently,this study discusses the research processes and application traits of intelligent algorithms in ecological optimization,prediction,and evaluation.The results demonstrate that:(1)In the field of ecological optimization,the Particle Swarm Optimization Algorithm(PSO)is effective for addressing problems involving continuous variable changes,while the Genetic Algorithm(GA)is more appropriate for scenarios characterized by discrete variables.Multi-agent Systems(MAS)exhibit advantages in solving sequential decision-making challenges.(2)In the domain of ecological prediction,deep learning algorithms are effective at handling high-dimensional and complex datasets,whereas Support Vector Machines(SVM)demonstrate efficiency and reliability when applied to smaller datasets.(3)Regarding evaluation models in ecological research,although individual models,such as the Decision Tree(DT),are prone to overfitting,they are well-suited for small datasets.In contrast,ensemble algorithms such as Random Forest(RF)and Gradient Boosting Decision Trees(GBDT)mitigate overfitting through their multi-tree structures,thereby enhancing generalization capacity and prediction accuracy.Furthermore,the advantages and limitations of the application of intelligent algorithms have been analyzed.Despite the potential of intelligent algorithms to optimize complex ecological systems and forecast future scenarios accurately,their application is constrained by s

关 键 词：生态学 智能算法 智能化优化 智能化预测 智能化评估 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] X171[自动化与计算机技术—控制科学与工程]