题名被动定位自适应卡尔曼滤波
作者钱东红
学位类别博士
答辩日期1991
授予单位中国科学院声学研究所
授予地点中国科学院声学研究所
中文摘要利用被动测距估计水目标的位置存在很大的困难,其原因在于时延估计不可能很精确,尤其在目标速度和航向角发生改变的情况下,用时延直接定位往往带来很大的偏差,甚至无法确定目标的位置,所以被动定位中都需要用一些后置处理的方法。卡尔曼滤波是比较好的一种后置处理方法,但目标距离、速度与时延之间的关系的非线性又给用线笥的卡尔曼结构带来困难。扩展的卡尔曼滤波可以解决非线性问题,但缺点是计算烦琐。本文采用两极滤波的新方法:第一步就是在使用非线性方程以前,直接对两个时延的差进行卡尔曼滤波,减少噪声造成的误差,因为两个时延的差在计算距离的近似公式中起着主要的作用。第二步,使用近似公式计算出距离和目标对本船的舷角,并进行第二次卡尔曼滤波,对距离和舷角滤波,便可得到较精确的目标参数,在整个滤波过程中,控制第一步滤波是个关键。由于卡尔曼算法中,增益系数是随指数衰减的,到达一定时间后增益系数非常小,而且几乎不变,而距离并不是线性变化的,尤其目标运动轨迹发生突变时,就会明显出现跟不上的现象。本文采用x~2检测的方法,对滤波过程的每一点进行检测,通过修正验前方差,及时增大增益系数,使滤波适应环境的变化,自动回到正确的位置上。此外,本文还对运动方程中的加速度项进行了估计,减少了近似公式中造成的偏差;滤波过程中对观测噪声的方差进行最小二乘估计,使滤波曲线得以平滑。两次卡尔曼滤波是在使用非线性方程前以及后进行的,只通过非线性方程的近似公式进行转换,既避免了扩展目测尔曼滤波的复杂计算又使估计结果达到较好的准确度。模拟实验以及实际数据都得到了较满意的结果。
英文摘要In the past several decades, a great deal of effort has been expended in the development of tracking system of underwater maneuvering targets. Although many systems have been developed, tracking problems continue to command a great deal of research effort. Because the time delay measurement is not accurate enough, it usually causes an obvious deviation by using direct localization. So a post-processing is helpful for passive localization. Kalman filter is a very good post-processing method. However, the nolinearity between time delay measurement and range estimate brings difficulty for the linear structure of Kalman filter. Extended Kalman filter may be a useful method but it is too complicated. In this paper, a two stage filter is proposed. The first stage is the filtering of the second difference between the time delays. The second stage is the ordinary filtering which is to range and bearing. It is neccessary to select a system which is suitable to an unpredictable model because the sea situation is very complicated. So the second improvement of this paper is to adopt Chi-square test for abrupt change of target. This point is also useful in the reduction of range deviation. Besides, the acceleration in the second stage is estimated and added to the system. In the whole filtering processing, the measurement noise variance is calculated by using the N most recent residuals. Fianally, computer simulation results and pratical sea data are shown which validate the elimination of extended Kalman filters in the measurement processing. This makes the system proposed very "robust" with respect to convergence characteristics in the presence of adverse target maneuvers.
语种中文
公开日期2011-05-07
页码39
内容类型学位论文
源URL[http://159.226.59.140/handle/311008/1166]  
专题声学研究所_声学所博硕士学位论文_1981-2009博硕士学位论文
推荐引用方式
GB/T 7714
钱东红. 被动定位自适应卡尔曼滤波[D]. 中国科学院声学研究所. 中国科学院声学研究所. 1991.
个性服务
查看访问统计
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。


©版权所有 ©2017 CSpace - Powered by CSpace