摘要
地理学正在经历“剧变”时代,愈发强调决策支持。地理决策过程中往往涉及众多因素,需要通盘考量不同因素、权衡利弊,并提出最优方案,是典型的多目标优化过程。数学领域的多目标优化算法因而在地理决策中具有广阔的应用潜力和重要意义,其最新进展是地理学中新方法、新工具的重要来源。本文综述了多目标优化领域最前沿、最流行的算法代表“非支配排序遗传系列算法”(Nondominated Sorting Genetic Algorithms,NSGAs),对其三代不同算法的原理、适用性进行对比,并综述了这些算法在地理决策领域的应用现状、改进方法、问题与局限。研究发现:在三代算法中,NSGA-Ⅱ因其在计算复杂性和使用场景方面的优势在地理决策中最为流行;NSGA-Ⅲ对建模要求较高,尚未得到广泛关注。在地理决策的各领域中,水资源管理领域是NSGA算法应用最多、最成熟的领域,该领域的问题建模和在NSGA算法中融入局部搜索的经验值得其它领域借鉴和推广;土地利用规划领域提出了较多的NSGA改进算法,为更好地融合NSGA算法进行地理决策树立了典范。未来研究中,可通过凝练行业共性问题、构建通用优化模型降低NSGA算法的应用门槛。建议通过深入结合局部搜索的方式提升NSGA算法的收敛速度,同时建议在地理过程模拟中深入融合NSGA等多目标优化算法。
The focus of geography is shifting from qualitative descriptions and quantitative analysis to support decision-making.The process of geographic decision-making usually involves multiple factors to consider and balance to achieve an optimal solution.It is a typical process of multi-objective optimization.Thus,multiobjective optimization algorithms from the field of mathematics are fundamental and have great potential to be applied in geographic decision-making.New algorithms of multi-objective optimization serve as an important source of new methods and tools for geography.This paper reviews a series of Nondominated Sorting Genetic Algorithms(NSGA-Ⅰ/Ⅱ/Ⅲ),which are among the cutting edge and most popular algorithms in the field of multiobjective optimization.This review summarizes the principles,applications,improvements,and problems of these NSGA algorithms.Our findings include:NSGA-Ⅱis the most popular algorithm among the series because of its low computational complexity and high usability;NSGA-Ⅲhas few applications in geographic decisionmaking for its sophisticated principles;currently,water resource management is the most successful field in applying the NSGA algorithms,and the experiences from this field are of use to others;and land use planning is the most successful field in improving the NSGA algorithms,making the NSGA algorithms more applicable to geographic decision-making.In the future,it is necessary to reduce the difficulty of applying the NSGA algorithms by summarizing typical issues in geographic decision-making and by developing user-friendly software tools for geographers.The efficiency of the NSGA algorithms can be further improved by coupling local searching strategies.It is also recommended to deeply incorporate the NSGA algorithms into the processes of geographic simulations.
作者
高培超
王昊煜
宋长青
程昌秀
沈石
GAO Peichao;WANG Haoyu;SONG Changqing;CHENG Changxiu;SHEN Shi(State Key Laboratory of Earth Surface Processes and Resource Ecology,Beijing Normal University,Beijing 100875,China;Faculty of Geographical Science,Beijing Normal University,Beijing 100875,China;National Tibetan Plateau Data Center,Beijing 100101,China)
出处
《地球信息科学学报》
CSCD
北大核心
2023年第1期25-39,共15页
Journal of Geo-information Science
基金
中国科学院“美丽中国生态文明建设科技工程”A类战略性先导科技专项(XDA23100303)
第二次青藏高原综合考察研究(2019QZKK0608)
国家自然科学基金(42171088、42171250)。
关键词
地理决策
多目标优化
非支配排序遗传系列算法
帕里托最优解
geographical decision
multi-objective optimization
Nondominated Sorting Genetic Algorithms
Pareto front
