摘要
非线性单纯形和模拟退火算法的结合形成了一种新的适合非凸集连续不受约束和受约束函数的全局最优化算法.静校正问题中的相位转移与模拟退火法有明显的相似之处.由非线性单纯形来产生系统的各个状态,在缓慢冷却过程中得到目标函数的全局最优解.所采用的目标函数是道集间的相关,解决了零空间现象.单纯形模拟退火(SIMPSA)算法应用于剩余静校正,收敛速度快,适于解决大剩余静校正.
An algorithm suitable for the globle optimization of nonconvex continuous unconstrained and constrained function is presented. The scheme combines the non-linear simplex and simulated annealing algorithms. The phase of statics is similar to the simulated annealing algorithms. The non-linear simplex creates the states of the system . During the course of annealing the objective function gets its global optimization result . The objective function we use is based on the coherence between neighboring CMP gathers . This can solve the null space . The SIMPSA is applied on the large residual statics . It brings faster convergence and is suitable to solve the large residual statics.
出处
《地球物理学进展》
CSCD
2004年第2期341-347,共7页
Progress in Geophysics
基金
中国科学院知识创新重大项目(编号KZCX1-SW-18KZCX1-Y01)
国家自然科学基金重大项目(编号49894190)资助.
关键词
单纯形
模拟退火
全局优化
零空间现象
剩余静校正
simplex, simulated annealing, global optimization, null space, residual statics,
