School of Earth and Ocean Sciences, University of Victoria, Victoria, British Columbia V8W 3P6, Canada.
J Acoust Soc Am. 2013 Apr;133(4):EL274-80. doi: 10.1121/1.4794931.
This letter develops a Bayesian approach to matched-field tracking of multiple acoustic sources in a poorly-known environment. Markov-chain Monte Carlo methods explicitly sample the posterior probability density over source locations and environmental parameters, while analytic maximum-likelihood solutions for complex source strengths and noise variance in terms of the explicit parameters allow these parameters to be sampled efficiently. This produces a time-ordered sequence of joint marginal probability distributions over source range and depth, from which optimal track estimates and uncertainties are extracted. Synthetic examples consider tracking a submerged source in the presence of a louder shallow interferer in an unknown environment.
这封信提出了一种贝叶斯方法，用于在环境未知的情况下对多个声源进行匹配场跟踪。马尔可夫链蒙特卡罗方法明确地对声源位置和环境参数的后验概率密度进行采样，而复杂声源强度和噪声方差的解析最大似然解可以用显式参数表示，从而有效地对这些参数进行采样。这会生成一个关于声源范围和深度的联合边缘概率分布的有序时间序列，从中提取最佳跟踪估计值和不确定性。合成示例考虑在未知环境中对一个处于较浅干扰器中的水下声源进行跟踪。
