摘要
为提高复合形法的寻优成功率,对初始复合形以及第一次出现“最坏点映射失败”之复形的顶点进行变异粒子群优化算法操作,搜索得到较优解,并替换掉当前复形中的一个顶点。利用最大复形中心距替换准则,产生新的复合形。对新的复形继续进行基本复合形法的计算,构成了基于变异粒子群优化的改进复合形法。对两个复杂土坡的最小安全系数的搜索表明,改进的复合形法具有较强的全局搜索能力。
In order to increase the successful rate of the original complex optimization algorithm, the mutation particle swarm optimization （MPSO） calculation would be introduced for all design points of the initial complex and the current complex while the mapping of the bad design point was unsuccessful firstly. One of the design points is replaced by a point with quasi-optimum object function value that has been gotten by the MPSO based on the substitution rule of maximum complex central distance; and a new complex has been formed. The original complex optimization calculation would be continued with the new complex until the stopping criterion. The improved complex method is applied to searching minimum safety factors in two complex soil slopes; and the result has shown that the method is much more likely to locate the true minimum.
出处
《岩土力学》
EI
CAS
CSCD
北大核心
2005年第9期1393-1398,共6页
Rock and Soil Mechanics
关键词
土坡稳定
安全系数
粒子群算法
复合形法
非均匀变异
slope stability
safety factor
particle swarm optimization
complex method
non-uniformity mutation
