摘要
针对传统多态电路进化设计算法有效性问题,提出了基于动态评价方法的多态进化设计算法。在进化算法适应度评价阶段,利用适应度评价扩展对于不同模式下的电路分别进行评价,采用比较选择进行电路最优结构配置,防止了潜在解的丢失。将多态门与普通门混合使用,进行了Multiplier/Sorter及Majority/Parity两种多态电路的进化设计实验。实验结果表明,与传统多态电路进化算法相比所提算法进化代数减少了31.2%-77.7%,成功概率提高了11%-52%,具有进化迭代次数少、成功概率高的优势,提高了算法有效性。
To deal with the problem of the effectiveness of the traditional evolutionary design algorithm for polymorphic circuits, a new design algorithm based on dynamic evaluation method is proposed. The fitness evaluation expansion is presented to separately evaluate the circuit under different modes, while the comparison and selection is used to configure the optimal structure of the circuit in the stage of fitness evaluation, thus the loss of potential solution is avoided. The evolutionary design experiments for Multiplier/Sorter and Majority/Parity circuits are conducted by combining the polymorphic gate with conventional gate. The experimental results show that compared with the conventional algorithm, the iterations of evolution is decreased by 31.2 %-77.7% and the success probability is increased by 11 %-- 52%. The proposed algorithm has the less iterations and the higher probability of success, thus a better effectiveness.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2012年第3期354-359,共6页
Journal of Nanjing University of Aeronautics & Astronautics
基金
南京理工大学自主科研专项计划(20110ZYTS028)资助项目
南京理工大学科研启动基金资助项目
关键词
多态电路
进化设计
动态评价
polymorphic circuits
evolutionary design
dynamic evaluation