基于演化的信赖域方法

基于演化的信赖域方法 Trust Region Algorithms Based on Evolution 钟守楠 ZHONG Shou\|nan 1 2 first-author 蔡晓芬 CAI Xiao\|fen 1 2 钟良 ZHONG Liang 1 2 武汉大学数学与统计学院 武汉大学数学与统计学院 School of Mathematics and Statistics, Wuhan University School of Mathematics and Statistics, Wuhan University 武汉大学数学与统计学院,湖北,武汉,430072 武汉大学数学与统计学院,湖北,武汉,430072 把全局搜索性能优良的演化算法与具有总体收敛性能的信赖域算法相融合形成局部随机搜索与全局确定性搜索相结合的演化信赖域方法 ,它具有适应性广 ,收敛性能好和收敛速度快的特点 ,为解决复杂的非线性优化问题提供了一种有效算法 ,并证明了算法的收敛性 A hybrid searching method consists of trust region algorithms with global convergence and evolutionary method with random searching function in local field are established. It is an algorithm that is random searching in local field and deterministic searching in global field. In order to solve nonlinear function optimization problems, an effective algorithm is provided. It possesses the characteristic of extensive adaptability, good convergence and fast convergence rate. And the proof of its convergence is given. 演化 信赖域 非线性 最优化 evolution trust region nonlinearity optimization O224 国家自然科学基金 （70 1710 16）;亚太运筹中心;中国科学院管理决策与信息系统开放研究室资助项目 2002-05-01 2021-04-01 5