在基于模拟退火算法的基础上提出了一种改进温度下降函数和自适应的生成邻域解的新型算法.该算法通过新提出的温度下降函数,使得在初始温度较高的时候下降较为平滑,同时在邻域解的生成过程中采用新的生成邻域解的方式,充分实现算法的全局性,克服传统模拟退火算法易陷入局部最优解的困境;同时在温度较低时候,平滑的温度下降方式也有利于进行充分的局部搜索,取得最优解.实验结果表明,与传统的模拟退火算法相比,提出的新型的模拟退火算法在三维片上网络的映射过程中,在功耗和收敛速度两个方面有显著的提升.
A new method to improve the declined function of temperature and the adaptive generation of neighborhood solution was proposed based on the simulated annealing algorithm( SA). The algorithm mades the decline of temperature in the higher initial temperature more smooth by the new function,and it adopted the new way of generating neighborhood solution,thus fully realized the global convergence. So it could overcome the plight of the local optimal solution. And in the condition of low temperature,it could sufficiently find the optimal solution by this function. The experimental results showed that the improved simulated annealing algorithm( ISA) significantly improved the performance of power consumption and the speed of convergence compared with the SA.