Application of Evolution Sequential Number Theoretic Optimization in Global Optimization  

序贯数论优化方法的改进及其在全局优化中的应用(英文)

作　　者：刘洪谦[1] 袁希钢[1] 方开泰[2] 

机构地区：[1]天津大学化工学院,天津300072 [2]香港浸会大学数学系

出　　处：《Transactions of Tianjin University》2002年第4期221-225,共5页天津大学学报（英文版）

基　　金：StateMajorBasicResearchDepartmentProgram(No. G2 0 0 0 0 2 62 3)

摘　　要：Synthesis of chemical processes is of non-convex and multi-modal. Deterministic strategies often fail to find global optimum within reasonable time scales. Stochastic methodologies generally approach global solution in probability. In recogniting the state of art status in the discipline, a new approach for global optimization of processes, based on sequential number theoretic optimization (SNTO), is proposed. In this approach, subspaces and feasible points are derived from uniformly scattered points, and iterations over passing the corner of local optimum are enhanced via parallel strategy. The efficiency of the approach proposed is verified by results obtained from various case studies.过程系统综合问题通常是非线性和多峰的 ,求解该类问题的确定性方法通常会陷入局部极值陷井 ,随机方法则只能以一定概率得到全局最优解 .基于此 ,提出一种并行序贯寻优 (SNTO)方法 .该方法中 ,可行点均匀散布在可行域内 ,并行搜索模式为小样本条件下获得全局近最优解提供了可能 .对测试函数的有效求解验证了该方法的有效性 .

关 键 词：global optimization sequential number theoretic optimization parallel optimization 

分 类 号：O224[理学—运筹学与控制论]