论文部分内容阅读
针对带全局约束条件的工作流可靠性计算问题,提出一种基于均匀多样性适应度函数的多子群协同进化算法,将工作流可靠性约束优化转化为双目标优化问题;为提高粒子在进化过程中的搜索能力,进化群体被分解为若干子群;综合考虑双目标优化问题的特点,设计了一种新颖实用的均匀多样性适应度函数,让各子群体在不同方向上协同搜索目标解;最后根据其适应度排序构造了基于非支配集合的全局最优解.仿真实验表明所提算法具有良好的效率,求得的最优解集全部满足约束条件,且分布和质量均优于基于非支配档案的混合离散粒子群算法.
Aiming at the problem of workflow reliability calculation with global constraints, a multi-subgroup co-evolutionary algorithm based on uniform diversity fitness function is proposed to convert the constraint of workflow reliability into a two-objective optimization problem. In order to improve the evolution of particles In the process of searching ability, the evolutionary population is decomposed into several subgroups. Considering the characteristics of the two-objective optimization problem, a novel and practical uniform diversity fitness function is designed to let each sub-population search the target solution in different directions Finally, the global optimal solution based on the non-dominated set is constructed according to its fitness ranking.The simulation results show that the proposed algorithm has good efficiency and the optimal solution set satisfies all the constraints, and the distribution and quality are better than those based on Mixed Discrete Particle Swarm Optimization for Non - dominated Files.