用差异进化算法进行海洋环境参数反演

提出了用差异进化算法进行浅海环境参数匹配场反演，利用仿真数据对算法的有效性进行了评估，并与遗传算法、模拟迟火算法进行了比较，最后采用亚洲海实验(ASIAEX 2001)数据验证了方法的可行性。

On Applying Differential Evolution Algorithm to Inversion of Ocean Acoustic Parameters

We found that GAs (Genetic Algorithms) and SAs (Simulated Annealing Algorithms) were not completely satisfactory when applied to the inversion of ocean acoustic parameters as these parameters differ greatly in sensitivity. DE (Differential Evolution) algorithm has proved to be very efficient; besides its good convergence properties and its suitability for parallelization, it is conceptually simple and easy to apply. Subsection 1.2 gives, according to different perturbation methods and different perturbed parameter vectors, five implementation forms for DE. Subsection 2.2 makes simulated inversions for three typical conditions with GA, SA, and DE; the comparison of results of simulated inversions given in Table 1 shows that DE can get a cost function that is lower than those obtained with GA and SA under the same convergence criterion. Employing DE algorithm and using ASIAEX2001 (Asian Seas International Acoustic Experiments) data measured in the East China Sea, section 3 inverts the environmental parameters and attains the precision of wideband matched acoustic field. Table 3 gives the results of inversion. Subsection 3.2 analyzes the inversion results in Table 3 and this analysis shows that DE algorithm is efficient.