CAJ | 학술논문

为了便于精准灌溉的田间实际操作和管理,采用改进的蚁群聚类算法进行灌溉管理分区研究.蚁群算法的离散性和并行性特点对于数据的特征聚类非常适用,然而当数据量较大时,蚁群聚类在系统循环过程中,对数据的搜索时间较长,计算量大,因此将初始聚类中心作为蚂蚁的初始食物源加以引导,减少蚂蚁行走的盲目性,以达到降低计算量、加快聚类的目的.该文以土壤物理特性作为数据源,在运用主成分分析消除初始指标相关性的基础上,采用改进的蚁群聚类对研究区进行精准灌溉管理分区的划分.将改进的蚁群聚类分区结果与传统的K-均值聚类进行比较,前者分区结果表现出土壤物理特性区内均一性更强、区间差异性更加显著的特点.改进的蚁群聚类分区结果表明,研究区可划分成2个灌溉管理分区,Ⅰ区土壤的田间持水率、饱和含水率和凋萎含水率均较Ⅱ区大,表明在相同的气候条件下,Ⅰ区的土壤耐旱能力较Ⅱ区强,分区结果可以为精准灌溉的分区管理提供参考依据和数据支持.
위료편우정준관개적전간실제조작화관리,채용개진적의군취류산법진행관개관리분구연구.의군산법적리산성화병행성특점대우수거적특정취류비상괄용,연이당수거량교대시,의군취류재계통순배과정중,대수거적수색시간교장,계산량대,인차장초시취류중심작위마의적초시식물원가이인도,감소마의행주적맹목성,이체도강저계산량、가쾌취류적목적.해문이토양물리특성작위수거원,재운용주성분분석소제초시지표상관성적기출상,채용개진적의군취류대연구구진행정준관개관리분구적화분.장개진적의군취류분구결과여전통적K-균치취류진행비교,전자분구결과표현출토양물리특성구내균일성경강、구간차이성경가현저적특점.개진적의군취류분구결과표명,연구구가화분성2개관개관리분구,Ⅰ구토양적전간지수솔、포화함수솔화조위함수솔균교Ⅱ구대,표명재상동적기후조건하,Ⅰ구적토양내한능력교Ⅱ구강,분구결과가이위정준관개적분구관리제공삼고의거화수거지지.
For more efficiently applying field operation and management of precision irrigation, an improved ant colony clustering algorithm was used to delineate irrigation management zones. Ant colony algorithm with the characteristics of discreteness and parallelism is applicable to data feature clustering. However when the data quantity is huge, ant colony clustering will take long time on data search and cause high computational complexity in the process of system circulation. Thus, for the purpose of decreasing computational complexity and accelerating clustering, initial clustering center was taken as the initial food source in the paper to guide ant colony to reduce the blindness of ant walking. Soil physical properties were taken as the data sources. After principal components analysis was used to eliminate correlations among initial indexes, improved ant colony clustering was performed to delineate site-specified irrigation management zones. According to the comparison of delineation management zones between improved ant colony clustering and K-means clustering, management zones delineated by the former showed the features that soil physical properties had stronger uniformity within the subzone and more significant difference between subzones. Delineation result based on the improved ant clustering indicated that the study area could be partitioned into two irrigation management zones. Soil field capacity, saturation moisture content and permanent wilting point in Zone Ⅰ were greater than those in Zone Ⅱ, which indicated that soil in Zone Ⅰ had stronger drought resistance than that in Zone Ⅱ under the same climate conditions. Delineation of irrigation management zones could provide references and data support for site-specified irrigation management.