不确定海洋环境中基于贝叶斯理论的声源运动参数估计方法

发布时间：2018-01-08 12:27

本文关键词：不确定海洋环境中基于贝叶斯理论的声源运动参数估计方法　出处：《物理学报》2016年16期 　论文类型：期刊论文

【摘要】：环境参数失配导致定位性能大幅度下降是匹配场定位所面临的难题之一.应用贝叶斯理论对环境聚焦,是当前解决该难题的研究热点.环境聚焦方法的实质是将未知环境参数和声源位置联合优化估计.然而,运动声源的位置时变性限制了观测时间长度和观测信息量,因此不得不利用很有限的观测信息来实现众多参数的估计.当航速较快或是环境信息的不确定性较大时,环境聚焦方法的效果迅速变差.借鉴卡尔曼滤波处理非平稳过程的参数估计思想,对航速较恒定的声源,本文将多个时刻的接收信号同时反演,引入能够描述声源位置随时间变化规律的时不变参数,以较少的时不变参数间接反演多个声源位置,从而有效降低待估参数维数.同时将当前估计结果作为下一次反演的先验信息,建立新的先验分布和代价函数,有效补偿个别异常数据,实现运动声源的连续定位.该方法在相同的环境不确定条件下,大幅度增加了观测时间和观测信息量,可以较好地改善环境聚焦方法的定位效果.
[Abstract]:One of the difficult problems in matching field localization is that the mismatch of environmental parameters leads to a great decline in localization performance. The Bayesian theory is used to focus on the environment. It is a hot topic to solve this problem. The essence of the environment focusing method is to optimize the unknown environmental parameters and the sound source location. The temporal variation of the position of moving sound source limits the length of observation time and the amount of information. Therefore, we have to use very limited observation information to realize the estimation of many parameters, when the speed of navigation is higher or the uncertainty of environmental information is greater. Using Kalman filter for parameter estimation of non-stationary process, the received signals at several times are inversed simultaneously for sound sources with constant speed. The time-invariant parameters which can describe the variation of sound source position with time are introduced, and multiple sound source positions are indirectly retrieved with fewer time-invariant parameters. At the same time, the current estimation result is regarded as the prior information of the next inversion, and a new prior distribution and cost function are established to compensate the individual abnormal data effectively. This method can greatly increase the observation time and the amount of observation information under the same environment uncertainty condition, and can improve the localization effect of the environment focusing method.
【作者单位】：山东科技大学测绘科学与工程学院;中国科学院声学研究所声场声信息国家重点实验室;海洋石油工程(青岛)有限公司;
【基金】：山东科技大学人才引进科研启动基金(批准号:2014RCJJ004) 测绘公益性行业科研专项经费(批准号:201512034) 国家自然科学基金(批准号:41506111,41376108)资助的课题~~
【分类号】：P733.2
【正文快照】： 1引言匹配场声源定位和跟踪是水声学研究中的经典问题,近年来逐渐成为国内外水声学者研究的一个焦点.匹配场定位是建立在对海洋声传播精确建模基础上的目标被动定位方法,它将实测水声信号与模拟声场信号相匹配来确定目标的深度和距离[1].其特点是对环境参数失配、阵列幅度相

【相似文献】