文献类型：期刊文献

中文题名：有向图基因表达式程序的电路演化模型

英文题名：Evolutionary Modeling of Circuits Using Graph-Based Gene Expression Programming

作者：陶砚蕴[1];林家骏[1];徐萃华[1]

机构：[1]华东理工大学自动化研究所,上海200237

年份：2010

卷号：22

期号：1

起止页码：94

中文期刊名：计算机辅助设计与图形学学报

外文期刊名：Journal of Computer-Aided Design & Computer Graphics

收录：CSTPCD;;EI(收录号：20100912741164);Scopus;北大核心:【北大核心2008】；CSCD:【CSCD2011_2012】；

语种：中文

中文关键词：数字电路;电路演化模型;基因表达式;遗传规划;中立性

外文关键词：digital circuit evolutionary design model of circuits gene expression genetic programming

摘要：为提高组合逻辑电路的进化速率和成功率,在基因表达式的基础上,结合图形结构提出了有向图基因表达式(GGEP)模型.其中基因表达式包含符号和连接两部分,通过基因显示型映射得到的连接非循环有向图可以很好地描述进化电路,最大正确输出且逻辑门数最少的电路为最优电路;提出了中立变异算法,使变异发生在基因表达式的非编码区域,产生电路进化的中立效果,并分析了该算法的复杂度,测试了中立对电路演化的影响.电路实验结果证明:GGEP模型比其他方法成功率高,演化速率快4～20倍;中立存在的GGEP模型的收敛速率和成功率都要比非中立变异的演化模型高出近一倍.
To improve the evolution rate and success rate of combination logical circuit,an evolutionary design modeling of circuits using graph-based gene expression programming is presented which combines gene expression with graph structure;the genotype of GGEP is composed of symbols and input connectivity.The genotype can be mapped to a directed acyclic graph that is used to describe a circuit.The optimum circuit has maximum correct outputs with shortest encoding.Mutations take place at non-coding area will have neutral effect refers to selection.The neutral mutation algorithm is proposed and analysis its complexity.The experimental results on two circuits indicated that the proposed approach has a better success rate and convergence rate is 4~20 times faster than other methods.Success rate and convergence rate of evolutionary model with neutral mutations doubled in comparison with non-neutral mutations.