基于自然码遗传算法的边坡稳定性聚类分析  

A clustering method for evaluation of slope stability based on genetic algorithm

作　　者：赵胜利[1] 刘燕[1] 周瑞林[2] 

机构地区：[1]河北农业大学城建学院,河北保定071001 [2]河北农业大学校园规划处,河北保定071001

出　　处：《四川建筑科学研究》2008年第5期124-127,共4页Sichuan Building Science

基　　金：国家自然科学基金(70371046);河北省科技厅资助项目(06547004D)

摘　　要：应用遗传算法解决边坡稳定性评价的聚类分析问题。针对问题特点,提出了基于自然数的编码方案并设计了相应的选择、交叉和变异算子。计算结果表明,该方法能以较大概率找到全局最优解并能有效克服常规聚类方法中对初始聚类中心敏感以及聚类结果与样本输入次序有关等缺点。A dynamic clustering method for evaluation of slope stability is developed based on genetic algorithm. By incorporating features of the problems discussed, the corresponding genetic operators such as selection strategy, crossover operator, and mutation operator are designed to promote global search. Analyzing results show that the method is easy to find global optimum solution and can resolve the defects in the ordinary dynamic clustering methods such as sensitivity to the original cluster center and clustering results depending on the order of input patterns.

关 键 词：遗传算法 边坡稳定性评价 聚类分析 

分 类 号：TU413.6[建筑科学—岩土工程]