基于猴王遗传算法的演化硬件  

Evolvable hardware based on Monkey-King Genetic Algorithm

作　　者：冉欢欢[1] 潘旭东[1] 田俊林[1] 

机构地区：[1]中国工程物理研究院应用电子学研究所,四川绵阳621999

出　　处：《太赫兹科学与电子信息学报》2015年第3期482-485,共4页Journal of Terahertz Science and Electronic Information Technology

基　　金：中国工程物理研究院发展基金资助项目(2013B0403074)

摘　　要：演化硬件作为新的硬件载体,具有自组织、自适应、自修复的能力。本文设计了一种由查找表(LUT)和D触发器构成的基本演化硬件单元,能够满足组合逻辑电路和时序逻辑电路的设计要求。同时设计了兼容组合逻辑电路和时序逻辑电路的适应度评估函数。利用该模型对基于遗传算法的演化硬件和基于猴王遗传算法的演化硬件参数进行了优化设计。分析表明遗传算法的变异概率vP取值应在0~0.01之间,交叉概率对算法的性能影响可以忽略;猴王遗传算法的保留概率R取值应在0~0.85之间,变异概率vP应在0~0.01之间。对比分析了遗传算法和猴王遗传算法的性能:基于猴王遗传算法演化硬件的运算量只有基于遗传算法演化硬件的1/4,且能得到更接近目标电路的电路结构。As a new hardware carrier, Evolvable Hardware(EH) features the capabilities of selforganizing, adaptive, and self-healing. A basic evolutionary hardware unit consisting of a Look Up Table(LUT) and a D flip-flop is proposed to meet the requirements of designing combinational logic circuit and sequential logic circuit. Fitness evaluation function is designed to be compatible with the combinational logic circuit and sequential logic circuit. Parameters of evolvable hardware based on Genetic Algorithm(GA) and Monkey-King Genetic Algorithm(MKGA) are optimized using this model. It is showed that GA mutation probability vP should be valued from 0 to 0.01, and the effects of crossover probability on GA performance can be ignored;while for MKGA, the reserved probability from 0 to 0.85,and mutation probability vP from 0 to 0.01 according to the analysis. The performances of GA and MKGA are compared: the computation amount of EH based on MKGA is only 1/4 of that based on GA while MKGA could get the circuit structure much closer to the target circuit.

关 键 词：遗传算法 猴王遗传算法 硬件演化 组合逻辑 时序逻辑 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TN791[自动化与计算机技术—控制科学与工程]