基于协同多目标差分进化的可逆逻辑综合方法  

Reversible logic synthesis algorithmbased on cooperative multi-objective differential evolution

作　　者：王旭 WANG Xu(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学机械与汽车工程学院,上海201620

出　　处：《上海工程技术大学学报》2023年第4期409-413,共5页Journal of Shanghai University of Engineering Science

摘　　要：可逆逻辑电路可避免因信息丢失而产生的热耗散,从而有望解决集成电路热耗问题.将可逆逻辑电路设计问题抽象为带有强约束的多目标优化问题,提出以协同多目标差分进化算法为核心的可逆逻辑综合方法.该方法基于种群自适应调节的差分进化算法,结合协同进化的多种群策略优化多目标,经种群选择及自适应调节策略和Paroto最优评估评价更新候选个体.通过求解经典电路测试集,验证了所提方法的可行性及有效性.与经典及基于启发式算法的可逆逻辑综合方法相比,该方法生成的电路性能更优.Reversible logic circuits can avoid thermal dissipation due to information loss so that it is possible to solve the thermal dissipation problem of integrated circuits.As reversible logic circuit synthesis problem was modeled as a multi-objective optimization problem,a reversible logic synthesis method was proposed based on a cooperative multi-objective differential evolution algorithm.Differential evolution algorithm with self-adaptive population resizing mechanism(SapsDE)was adopted as the basis and combined with the coevolution algorithm based multiple population strategy for multiple objectives.Meanwhile,the population updating scheme and the fitness evaluation strategy based on Pareto-optimal were employed to update the candidate individuals.The synthesis method tested on a suite of benchmark functions is feasible and effective.Compared with classical and heuristic synthesis methods,the circuits generated by the proposed synthesis method have better performance.

关 键 词：可逆逻辑电路 差分进化 协同进化 多目标 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]