基于FPGA的遗传算法流水线设计与实现

摘要

针对软件实现遗传算法运行速度过慢的问题,设计一种基于FPGA的遗传算法流水线.硬件系统采用完全流水线结构,划分为选择、交叉、变异、适应度计算4个流水单元.在Cyclone系列芯片上进行实现,测试结果表明,基于硬件实现的遗传算法的运行速度比用软件实现快3个数量级.%Aiming at the problem that it is too slow to run genetic algorithm with software, this paper designs a pipelining of FPGA-based genetic algorithm. The system is implemented by means of a four-cascade full pipeline, i.e., dividing the system into reproduction, crossover, mutation and calculating fitness pipeline cells. The whole design is implemented in the Cyclone serial chips, and experimental result shows that the hardware implementation runs over three orders of magnitude faster than usual software implementation.