一种动态邻居元胞遗传算法  

Dynamic Neighbour Cellular Genetic Algorithm

作　　者：吴佳俊[1] 王帮峰[1,2] 芦吉云[2] 

机构地区：[1]南京航空航天大学机械结构力学及控制国家重点实验室,江苏南京210016 [2]南京航空航天大学民航学院,江苏南京210016

出　　处：《计算机仿真》2013年第1期351-355,共5页Computer Simulation

基　　金：国家自然科学基金(51075207);航空基金(2011ZA52013)

摘　　要：研究一种改进的元胞遗传算法。将遗传算法中的个体适应度和元胞自动机中的邻居定义做了结合,提出基于元胞间距离以及元胞个体适应度的"影响力算子",并作为算子中心元胞判断邻居的依据,从而形成改进算法,并对改进算法的基本性能的进行了两组定量分析,一是影响力算子对选择压和多样化损失的控制,另一部分是将算法与改良后传统元胞遗传算法做了对比测试。结果显示,即便使用最朴素的影响力算子而且不采用其它优化手段的情况下,算法依然能对选择压和多样化损失进行有效地控制,并且相较于使用了最优个体保持和小范围竞争择优的传统元胞遗传算法收敛率提高了约10%。This algorithm is an improvement to Cellular Genetic Algorithm（CGA）. An operator called ＂Influence＂ was used to choose the neighbours of each cell in the area. There are two factors in the operator： distance between candidate cell to central cell and the fitness of each candidate. Obviously, the neighbours of every cell changes constantly in evolution process. So it is called Dynamic Neighbour Cellular Genetic Algorithm（DN-CGA）. In order to quantitatively analyse the effectiveness of the algorithm, two groups of examples were processed. The ＂Selection pressure＂ control and ＂loss of diversity＂ control were estimated in the first group, and the second group was about lateral comparison between this algorithm and optimized traditional CGA. The results shown that even with the simplest ＂Influence operator＇, DN-CGA can control the ＂selection pressure＂ and ＂loss of diversity＂ effectively, and raised about 10% convergence rate than traditional CGA in the tests.

关 键 词：遗传算法 元胞自动机 影响力算子 邻居 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]