差分进化算法求解二阶段直觉模糊非线性规划  

Differential Evolution Algorithm for Two-stage Intuitionistic Fuzzy Nonlinear Programming

作　　者：徐小来[1] 雷英杰[1] 戴文义[1] 

机构地区：[1]空军工程大学导弹学院

出　　处：《系统仿真学报》2009年第17期5384-5387,共4页Journal of System Simulation

基　　金：国家自然科学基金(60773209);陕西省自然科学基金(2006F18)

摘　　要：针对Plamen直觉模糊规划模型需要同时考虑目标和约束函数的拒绝和满足程度,算法的计算量相当于模糊规划两倍的问题,提出了二阶段直觉模糊规划模型。即前期只考虑目标和约束函数的拒绝程度,使最优值集中在全局最优值附近,后期只考虑目标和约束函数的满足程度,使最优值靠近全局最优值,因此,算法的计算量仅为Plamen直觉模糊规划模型的一半。并用差分进化算法进行求解,根据前、后两阶段的特点,分别采用DE/rand/1和自适应变异算子。最后,通过Benchmarks测试函数验证了方法的有效性和稳定性。For Plamen ＇s intuitionistic fuzzy programming model needs to consider the degrees of rejection of objective and of constraints together with the degrees of satisfaction, its arithmetic complexity is twice as fuzzy programming. A two-stage intuitionistic fuzzy nonlinear programming model was proposed. In the first stage, only the degrees of rejection of objective and constraints were considered, which made minimums concentrate around the global minimum. In the second stage, only the degrees of satisfaction of objective and constraints were considered, which made minimums move towards global minimum. So its arithmetic calculation size is only half of Plamen＇s intuitionistic fuzzy programming model Then, two-stage intuitionistic fuzzy nonlinear programming was resolved by differential evolution algorithm, DE/rand/1 and adaptive mutation operators were used in two stages separately. At last, Benchmarks test functions validate stability and validity of the model.

关 键 词：直觉模糊集 模糊规划 非线性规划 差分进化算法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]