云南省水安全区域类型识别TSA-PP模型及应用

为科学识别云南省16个州市的水安全区域类型,提出了树-种算法(TSA)-投影寻踪(PP)识别模型。选取了4个典型测试函数对TSA进行仿真验证,并将验证结果与人工蜂群(ABC)算法、布谷鸟搜索(CS)算法等6种算法的仿真结果进行了对比。以云南省为研究对象从水资源条件、经济社会条件和水环境条件中遴选出了27个指标构建区域水安全类型识别指标体系和分级标准,在各分级标准阈值间采用随机内插的方法生成样本;同时,构建了基于水资源、经济社会和水环境条件的投影指标函数,并分别采用TSA搜索最优投影向量,计算云南省内各州市的综合投影值以及各分级标准阈值的投影值,然后利用分级标准阈值投影值对各行政区的水安全区域类型进行识别。结果表明:TSA寻优精度优于ABC、CS等6种算法,具有较好的收敛精度、极值寻优能力和收敛稳健性能。TSA-PP模型对云南省各州市的水安全类型识别结果为:昆明市、玉溪市为"中度缺水-较发达-中等"型;楚雄州、大理州为"严重缺水-中度发达-较差"型;西双版纳、迪庆州为"中度缺水-中度发达-中等"型;丽江市为"严重缺水-中度发达-中等"型;德宏州为"轻度缺水-中度发达-中等"型;怒江州为"轻度缺水-中度发达-较差"型;其他州市被识别为"中度缺水-中度发达-较差"型。

TAS-PP model and its applications in identifying water security zone type in Yunnan Province

TSA-PP identification model (tree-seed algorithm and projection pursuit) was proposed to scientifically identify the water security zone of sixteen prefectures in Yunnan Province. We selected four typical test functions to simulate TSA, and compared it with the results obtained by other six algorithms including ABC algorithm and CS algorithm. We chose twenty-seven indicators from water resources condition, economic and social conditions and water environment condition to construct the regional water safety type identification index system and grading standard. Random interpolation method was used to get samples between the thresholds of each classification standards; meanwhile, based on the three aforementioned conditions, projection indicator function was constructed. TSA was used to search the optimal projection vector. The comprehensive projection value of each prefecture and the threshold value of each grading standard were calculated. The graded standard threshold projection value was employed to identify the water area of each prefecture. The results show that TSA is superior to ABC and CS algorithm, and it has good abilities in convergence accuracy, extreme value search and convergence performance. The results (denoted by “water stress condition- economic/social condition- water environment condition) obtained by TSA-PP model are as follows: Kumming and Yuxi are “moderate-developed-moderate” type; Chuxiong and Dali prefectures are “severe-moderate developed-poor” type; Xishuangbanna and Diqing prefectures are identified as “moderate-moderate developed-moderate” type; Lijiang is “severe-moderate-moderate” type; Dehong prefecture is “mild-moderate developed-moderate” type; Nujiang prefecture is “mild-moderate-poor” type; other prefectures are “moderate-moderate developed-poor” type.