自适应演化硬件系统中遗传算法内核设计

自适应演化硬件系统中遗传算法内核设计 Genetic Algorithm Kernel Design for EHW-Based Self-Adaptive Systems 时晨 SHI Chen 1 first-author 黄士坦 HUANG Shitan 1 乔永强 QIAO Yongqiang 2 西安微电子技术研究所 西安微电子技术研究所 Xi’an Microelectronics Institute Xi’an Microelectronics Institute 中国航天时代电子公司研究院 中国航天时代电子公司研究院 Academy of China Aerospace Times Electronics Corporation Academy of China Aerospace Times Electronics Corporation 根据基于演化的自适应系统的体系架构,将与具体应用相对独立的遗传算法操作部分抽取出来,结合算法的操作流程,设计了一个与应用无关的、能够独立完成遗传算法操作的GAK（遗传算法内核）.采用VHDL（超高速集成电路硬件描述语言）语言描述的状态机来实现该设计,其中种群规模、染色体长度等参数均可配置.采用Xilinx的Virtex V1000FG680、速度等级为—6的FPGA（现场可编程门阵列）来实现GAK,其中染色体长度为8位、适应度值为16位、种群规模为256,此时该设计的综合结果为:占用资源为485个片（slice）,时钟频率为48.51MHz.进而在GAK上仿真求解了多项式函数的极值优化问题,从而验证了设计的有效性. In this paper iurther analysis is essential to the composition of EHW-based Adaptive Systems, which will be divided the system into two parts: genetic algorithm kernel for a genetic algorithm operation, as well as application-specific fitness needs assessment calculator. Genetic algorithms use VHDL （very high speed IC description language） to implement state machine, in which to configure population size, chromosome length, and all other parameters. When the chromosome length is eight, the fitness value of 16, and population size for the 256 individual circumstances, the synthesis results of GAK（genetic algorithm kernel） targeted in Xilinx Virtex V1000FG680, -6 speed FPGA（field programmable gate arra）: used 485 slice and the max clock frequency is 48. 51 MHz. And the optimization of extreme polynomial function showed the effectiveness of design. 演化硬件 自适应系统 遗传算法 遗传算法内核 evolvable hardware（EHW） self-adaptive systems genetic algorithm genetic algorithm kernel TP18 国防科学技术工业委员会民用航天项目（C5220060318） 2008-01-01 2021-04-01 1