基于水贫困指数及MSCS-SVM模型的云南省水安全评价

为科学评价云南省近10年水安全状况,利用多策略布谷鸟搜索(MSCS)算法搜寻支持向量机(SVM)最佳学习参数,提出MSCS-SVM模型。首先选取6个典型测试函数对CMCS算法进行仿真验证,并与布谷鸟搜索(CS)算法的仿真结果进行对比;其次基于国际上通用的水贫困指数构建区域水安全评价指标体系和分级标准,采用在分级标准阈值间随机内插和随机选取的方法构造训练样本及检验样本,以对MSCS-SVM模型进行训练及检验;最后利用MSCS-SVM模型对云南省近10年及2020年水安全状况进行评价分析。结果表明:MSCS算法寻优精度远优于CS算法,表现出较好的收敛速度、极值寻优能力和全局搜索性能。利用MSCS-SVM模型对云南省2006年水安全进行评价,结果为不安全,2007—2015年评价结果为临界,2020年评价结果为安全,近10年水安全评价结果随时间呈提升趋势,且提升趋势显著。

Evaluation of water safety in Yunnan Province based on water poverty index and MSCS-SVM model

In order to evaluate the water security situation of Yunnan Province in the past 10 years, by using the multi-strategy cuckoo search(MSCS)algorithm to search the optimum learning parameters of the support vector machine(SVM), a MSCS-SVM evaluation model for water safety is put forward. Firstly, six typical test functions are selected to simulate the CMCS algorithm, and the results are compared with the simulated ones of the cuckoo search(CS)algorithm. Secondly, the regional evaluation index system and grading standard of water safety are established based on the international common water poverty index. The MSCS-SVM model is trained and tested by constructing the training samples and the test samples by random interpolation and randomly selected thresholds. Finally, the MSCS-SVM model is used to evaluate the water security situations of Yunnan Province in recent 10 years and 2020. The results show that the optimization accuracy of the MSCS algorithm is much better than that of the CS algorithm, and it has a better convergence speed, extreme search capability and global search performance. By using the MSCS-SVM model, the evaluated results of water security in Yunnan Province in 2006 are unsafe, they are rated to be critical from 2007 to 2015, and they are safe in 2020. The water safety of Yunnan Province is increasing in the past 10 years, and the increasing trend is significant.