梯度退火混合优化算法

Gradient annealing hybrid optimal algorithm

通过对梯度法与模拟退火算法优缺点的分析,提出了一种梯度退火新型混合全局优化算法。该算法利用梯度法的快速寻优特性得到某一局部极值,然后采用模拟退火算法的全局搜索寻优能力跳出该局部极值,经过反复混合迭代最终获得全局最优解。仿真实验表明,该新型混合优化算法显著提高了求解全局优化问题的计算效率。

Based on analysis of merits and defects of gradient method and simulated annealing algorithm, a new hybrid optimal algorithm, which is combined by simulated annealing algorithm and gradient method, is proposed. Firstly ,the new algorithm used quickness searching of gradient method to obtain a local minimum,then by utilizing the abilities of global searching of simulated annealing algorithm,it escaped from trapping this local minimum. At last, the global minimum was achieved through iterative computation. Simulation experiments showed that this new hybrid algorithm can significantly improve the computational efficiency of solving global optimization problems.