摘要
针对三值FPRM电路面积与功耗综合优化问题,提出一种基于差分非支配排序遗传算法(Differential Non-dominated Sort Genetic Algorithm Ⅱ,DNSGA-Ⅱ)的最佳极性搜索方案.首先在DNSGA-Ⅱ算法中,随机抽取种群个体进行高斯变异而产生变异群体.从Pareto非劣解集和变异群体中抽取父代进行二项式交叉产生子代群体,从而维持算法的多样性.然后,结合DNSGA-Ⅱ算法与三值FPRM电路极性转换技术和低功耗技术,搜索电路面积与功耗的最佳极性.最后对MCNC Benchmark电路进行测试,与GA和NSGA-Ⅱ算法搜索到的结果相比,DNSGA-Ⅱ算法获取的最佳极性电路功耗平均减小19.53%和15.08%,面积平均节省9.01%和6.05%.
Taking the minimum area and power of ternary FPRM circuits as objectives,a Differential Non-dominated Sort Genetic Algorithm Ⅱ（DNSGA-Ⅱ） is proposed.Firstly,in DNSGA-Ⅱ algorithm,the differential evolution strategy is introduced.The individuals are randomly selected from population to produce mutation groups by Gaussian mutation.And some individuals are selected from the mutation group and Pareto non-inferior group as the parents to generate offspring group by binomial crossover,which maintain the diversity of the algorithm.Secondly,combined with the polarity conversion technique and the low power technique,DNSGA-Ⅱ will search the best polarity from the ternary FPRM circuits to optimize the circuits area and power.Finally,the PLA format MCNC Benchmark circuits are put to test to verify the effectiveness of DNSGA-Ⅱ,and the result shows that the average saving of power and area achieves 19.53%,15.08% and 9.01%,6.05% in comparison with GA and NSGA-Ⅱ respectively.
作者
王铭波
汪鹏君
符强
张会红
WANG Ming-bo;WANG Peng-jun;FU Qiang;ZHANG Hui-hong(Institute of Circuits and Systems,Ningbo University,Ningbo 315211,China;College of Science ＆ Technology,Ningbo University,Ningbo 315212,China)
出处
《宁波大学学报（理工版）》
CAS
2018年第5期40-44,共5页
Journal of Ningbo University：Natural Science and Engineering Edition
基金
国家自然科学基金(61474068
61306041)
浙江省公益性技术应用研究计划项目(2016C31078)
关键词
差分非支配排序遗传算法
三值FPRM电路
极性搜索
低功耗技术
differential non-dominated sort genetic algorithm Ⅱ
ternary FPRM circuit
polarity search
low power technique
