基于整数编码的分布式估计单元重组算法

Distributed estimation of reorganized call algorithms based on integral number coding

分布式估计算法(estimation of distribution algorithms,EDAs)源于遗传算法,是一种对群体采用宏观方法建模和模拟的新型优化算法。本文根据动态重组制造单元的一般原则,针对制造单元重组问题,提出了基于整数编码的分布式估计制造单元重组算法,通过概率图模型取代传统的交叉、变异等遗传算法操作,并创新的采用轮盘赌方法将EDAs的编码由二进制数改进为整数,降低了编码的长度,开拓了其应用范围。该算法根据两个定量指标,将动态逻辑单元重组问题转化成一个简单的多目标组合优化问题模型,并能真实反映单元重组的本质要求。最后给出实例验证。

Estimation of Distribution Algorithms （EDAs） is a class of novel optimization algorithm stemed from the Genetic Algorithms, which used macroscopic approach to model and simulate the colony. According to the principle of dynamica reorganized manufacturing cells, this paper advanced the estimation of distributed algorithm by introducing a new paradigm of evolutionary computation without using conventional evolutionary operators of Genetic Algorithms such as crossover and mutation. With the Roulette method the coding of the EDAs used to be binary code was improved onto integral number code which reduced the length of individuals and explored the scope of EDAs application. According to two quantitative indexes, this algorithm transformed the problems of dynamic logic reorganized cell into a simple combination of multi-objective optimization problem, which reflects the essence of dynamic logic of these cells. An example was given to demonstrate the algorithm at last.