遗传退火算法在软硬件划分中的应用

Application of Genetic Annealing Algorithm in Software-hardware Partition

针对基于IP核的软硬件划分组合问题,提出一种改进的自适应最优保存的遗传退火算法。将最优保存遗传算法和模拟退火算法相结合,把循环策略应用到混合算法中,并在变异概率中引入自适应的概率变化,自适应地保存最优个体,有效地解决了这2种算法的早熟现象和时间问题。仿真实验表明该算法有效地解决了软硬件划分问题,具有较强的搜索和跳出局部最优的能力。

To resolve the software-hardware partition problem that based on IP core, an improved algorithm that adopts adaptive optimal preservation is proposed. This algorithm combines the Maintaining Optima Saved Genetic Algorithm（MOSGA） and Simulated Anneaiing（SA） algorithm, introduces the cycle strategy and the adaptive probability changes to the mutation, and adaptive optimal preservation, which effectively solutes the matter of premature and time. The simulation shows that the algorithm is effective to the problem of software-hardware partition, and has strong ability of searching and getting out of the local optimum.