用模拟退火差异进化算法进行匹配场反演

分析了模拟退火算法与差异进化算法单独应用于匹配场反演时的局限性，构造一种结合两种算法优点的混合算法。模拟退火算法对差异进化算法的种群进行扰动，保留了差异进化算法的并行特性；在差异进化算法中引入模拟退火算法的重要性抽样思想，增强其逃离局部最优的能力，差异进化算法对目标函数梯度信息的利用仍然保留。仿真研究表明，混合算法在寻优时间与效率上达到了较好的平衡。

Matched Field Inversion via Hybrid Simulated Annealing-Differential Evolution Algorithm

A hybrid optimization algorithm that combines advantages of two global optimization algorithms (simulated annealing algorithm and differential evolution algorithm) is developed and applied to the problem of determining seabed properties by minimizing the mismatch between measured and modeled acoustic fields, since there are a lot of problems in both of the algorithms. In the proposed hybrid SADE algorithm the population of DE is perturbed by SA, while the parallel property of DE is preserved. The Metropolis algorithm of SA is introduced into DE, which intensifies its ability to escape from local minima. Furthermore, the use of gradient information of DE can improve the search of parameter space more efficiently. Examples based on synthetic data show that the proposed hybrid algorithm works well.