基于花授粉算法的结构面自适应精细分组研究  

Study on adaptive fine grouping of structural planes based on flower pollination algorithm

作　　者：王资平 林锋[1] 丁秀美[1] 黄星凯 石广源 王卫[1] WANG Ziping;LIN Feng;DING Xiumei;HUANG Xingkai;SHI Guangyuan;WANG Wei(State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu 610059, China)

机构地区：[1]地质灾害防治与地质环境保护国家重点实验室(成都理工大学),成都610059

出　　处：《成都理工大学学报（自然科学版）》2022年第2期219-224,共6页Journal of Chengdu University of Technology: Science & Technology Edition

基　　金：地质灾害防治与地质环境保护国家重点实验室自主研究课题(SKLGP2019Z011);国家创新研究群体科学基金项目(41521002)。

摘　　要：岩体结构特征量化分析的前提和基础是对大量结构面快速合理分组,需要提出更高效的结构面自适应分组方法。引入花授粉算法,将结构面精细分组问题概化为多目标组合优化求解问题;基于结构面方位相似度定义了分组目标函数,选择Silhouette指标来判别最佳分组数,采用花授粉算法进行自适应寻优求解。对某大型水电工程坝基岩体中结构面进行分组应用表明,采用花授粉算法,中陡倾角结构面的最优分组结果与极点等密图法高度一致,并能清晰得出缓倾角结构面的分组结果,且与现场判断一致。The premise and basis of quantitative analysis of rock mass structural characteristics is to group a large number of structural planes quickly and reasonably,and to propose a more efficient adaptive grouping method of structural planes.In this paper,the flower pollination algorithm(FPA)is introduced to generalize the structural plane fine grouping problem into a multi-objective combinatorial optimization problem.The grouping objective function is defined based on the azimuth similarity of the structural planes,Silhouette index is selected to identify the optimal grouping number,and the flower pollination algorithm is used for adaptive optimization.The practical application of grouping of structural planes in dam foundation rock mass of a large hydropower project shows that the optimal grouping result of medium-steep tilting angle structural planes by using flower pollination algorithm is highly consistent with the pole isometric graph method,and the grouping result of low-dipping angle structural planes can be clearly obtained,which is consistent with the field judgment.

关 键 词：花授粉算法 自适应分组 Silhouette指标 聚类分析 

分 类 号：P642[天文地球—工程地质学]