基于TS-SVM模型的水安全评价  

Water safety assessment based on TS-SVM model

作　　者：汪嘉杨[1,2] 王文圣[2] 李祚泳[1,2] 张碧[1] 

机构地区：[1]成都信息工程学院,四川成都610041 [2]四川大学水电学院,四川成都610065

出　　处：《水资源保护》2010年第2期1-4,9,共5页Water Resources Protection

基　　金：科技基础性工作专项(2007FY140900);国家自然科学基金(50779042;50739002);成都信息工程学院科研基金(CRF200810)

摘　　要：核函数参数和误差惩罚因子的选择对支持向量机模型(SVM)的精度有较大影响。为充分发挥SVM的性能,提出基于禁忌搜索算法优化的支持向量机模型(TS-SVM)。将TS-SVM应用于山东省水安全指数计算,得到各个水安全利用方案的评价指数值和安全等级。对比分析表明,TS-SVM具有较强的学习能力,能给出合理的评价结果,是一种有效的评价模型。The selection of the kernel function parameter and error penalty factor affected the precision of the support vector machine （SVM） significantly. In order to give full play to the SVM model, a TS-SVM model in which parameters were optimized by a taboo search algorithm （TS） was pmposod. By using this model to calculate the evaluation indexes of the water safety of Shandong Province, safety indexes and safety grades of every utilization project were obtained. The results show that the TS-SVM model, which has strong learning ability, can yield rational results and is a practicable model for water safety assessment.

关 键 词：支持向量机 禁忌搜索算法 水安全评价 安全指数 

分 类 号：TV213.4[水利工程—水文学及水资源]