匹配场反演后验概率分析

A posteriori Probability Analysis for Matched Field Inversion

构造了一种基于并行优化算法的后验概率分析算法,用于对匹配场反演结果进行定量的概率分析。应用并行优化算法的强收敛特性,解决了统计计算中常出现的收敛标准难以确定的问题;应用优化算法高度并行、种群个体收敛高度一致的特性,较好的解决了高维统计计算中样本数量与计算时间、样本数量与样本质量之间难以协调的问题。匹配场反演仿真算例表明,该算法能有效对反演结果进行概率分析。

A posteriori probability analysis method was proposed based on a parallel optimization algorithm, and applied to quantitative analysis of matched inversion results. The frequently appeared the uncertain of convergence criterion was resolved. The problems of harmony between sample number and calculation time, as well as between sample number and sample quality in higher order statistics calculation were achieved. Examples based on synthetic data show that posteriori probability analysis for matched field inversion can be conducted effectively by the proposed parallel optimization algorithm.