SIA OpenIR  > 海洋信息技术装备中心
强杂波环境水下多目标跟踪方法关键技术研究
其他题名Research on key technology of underwater multi-target tracking in the heavy clutter environment
李冬冬1,2
导师林扬
分类号U675.7
关键词强杂波环境 大误差量测 目标跟踪性能评估方法 系统因素 非线性因素
索取号U675.7/L31/2018
页数115页
学位专业机械电子工程
学位名称博士
2018-05-21
学位授予单位中国科学院沈阳自动化研究所
学位授予地点沈阳
作者部门海洋信息技术装备中心
摘要

本文的研究工作依托于中国科学院重点项目(No.KGFZD-125-014),在查阅、理解、剖析、总结大量国内外水下多目标跟踪领域、参数估计领域等相关研究成果的基础上,结合研究室项目对水下多目标跟踪任务的需求,针对水下特殊环境(强杂波干扰和大误差量测环境),对强杂波环境水下多目标跟踪相关理论和技术进行深入研究,主要研究内容如下:1. 如何量化目标跟踪环境;2. 如何评估目标跟踪性能;3. 如何提高水下多目标跟踪性能。针对第一个问题,如何量化目标跟踪环境。现阶段大多数目标跟踪方法都是针对具体环境而言,但没有从数学角度量化阐述目标跟踪环境,此给目标跟踪方法的泛化理解带来困难。只有量化目标跟踪环境之后,才能更准确地理解目标跟踪环境,进而有助于选择和改进目标跟踪方法。针对第二个问题,如何评估目标跟踪性能。现如今,学术界尚未有人提出令人满意的评估目标跟踪性能的方法。现有的目标跟踪性能评估方法以及目标跟踪方法普遍存在以下问题:1. 目标跟踪方法缺乏有意义的横向对比;2. 目标跟踪方法缺乏有深度的纵向对比;3. 现有的目标跟踪性能评估方法角度比较片面;4. 现有的目标跟踪性能评估方法是一种后验评估方法。目标跟踪方法缺乏有意义的横向对比,是指不恰当地将论文所提出的针对具体环境下的改进方法与其它通用的方法进行对比。在某种程度上讲,这种对比效果是不具备参考价值的。目标跟踪方法缺乏有深度的纵向对比,是指目标跟踪方法是否仍有进步的空间。到目前为止,本人所查阅的所有文献中,少有涉及探讨目标跟踪性能上限的问题。现有的目标跟踪性能评估方法角度比较片面,是指分析与对比不同方法性能优劣的角度单一。分析与对比不同的目标跟踪方法不仅仅需比较方法的综合性能,还需具体分析哪些因素导致了目标跟踪性能的恶化,哪些因素是主要因素,哪些因素是次要因素。应该针对主要因素对目标跟踪方法进行改进,其次是针对次要因素。如果没有完备而且详细的影响因素评估,论文所提出的方法就有些盲目性,无法解释清楚所提方法的意义以及其泛化应用能力。现有的目标跟踪性能评估方法是一种后验评估方法。目前常用的目标跟踪性能评估方法是均方根误差,其是在某一具体场景中,通过仿真获得目标的估计值和真实值之间的均方根误差,即目标跟踪性能评估值。其优点是直观和置信度高。但是均方根误差是一种带有盲目性质的后验性能评估方法。如果有一种先验性能评估方法,能够提前预知目标跟踪方法的整体性能评估值和各影响因素方面的性能评估值,则可以提前指导筛选以及改进目标跟踪方法,有效地克服盲目性。针对第三个问题,提高水下多目标跟踪性能。任何目标跟踪方法的初衷均是快速准确地获得目标状态的实时值。针对水下特殊环境,首先清楚当前目标跟踪方法的优势点和劣势点以及环境特点,即综合性能评估和各影响因素方面的性能评估,抓住、聚焦与解决主要矛盾,进而针对性地提出解决方案,获得更好的目标跟踪性能。此就将前两个问题和最后一个问题联系起来。本文针对特殊性的水下环境,从多个影响目标跟踪性能的因素方面分别提出了改进方法,如航迹起始方法、数据关联方法、轨迹终结方法和水下高精度导航方法,从而提高目标跟踪整体性能。

其他摘要

This thesis depends on the Key Research Program of the Chinese Academy of Sciences (No. KGFZD-125-014). On the basis of investigating, understanding, analyzing, summarizing target tracking related research achievements (multi-target tracking and parameter estimating field) at home and abroad, combined with the actual function requirement of the program, in order to deal with underwater special environment (heavy clutter and large measurement error), we widely and deeply research the conception and technology of underwater multi-target tracking in the heavy clutter environment. Specific contents are as follows: 1. How to quantify the target tracking environment. 2. How to evaluate the performance of target tracking method. 3. How to improve the performance of target tracking method. For the first content, quantify the target tracking environment. It means most of the target tracking methods is aimed at a specific environment, but not form the mathematical perspective, which qualifying the environment of target tracking. It brings understanding difficulty for generalizing target tracking method. After qualifying the target tracking environment, then we can accurately understand the environment, and it is used to select and modify target tracking methods. For the second content, evaluate the performance of target tracking method. Currently, there is no perfect and satisfactory performance evaluation method for target tracking method. There are some shortcomings for the existing performance evaluation method and target tracking methods: 1. Target tracking method is lack of meaning comparison comparing with other target tracking methods. 2. Target tracking method is lack of meaning comparison as for the upper bound of target tracking performance. 3. The existing performance evaluation method of target tracking is one-sided. 4. The existing performance evaluation method of target tracking is a posteriori assessment method. Target tracking method is lack of meaning comparison comparing with other target tracking methods. It means we should not compare a new method with a common method as for the specifically environments. In some extent, the comparing has no referent value. Target tracking method is lack of meaning comparison as for the upper bound of target tracking performance. It means that existing performance evaluation cannot state the upper bound of a target tracking method, which indicates how many improvement and work to do. So far, none of the papers published I have reviewed have addressed the issue about the upper bound of target tracking performance. The existing performance evaluation method of target tracking is one-sided. It means that we only compare and analyze the target tracking performance in a single point of view. We should analyze the target tracking from multiple perspectives, including comprehensive performance and subdivided performance. We should find out which factors is the main factor leading to decreasing of target tracking performance, and which factors is the secondary factors that can be ignore. It should concentrate attention and improvement measure on the main factor, and deal with the secondary factors after solving the main factor. If there is no complete and detailed performance evaluation of the influencing factors, the method proposed is somewhat blind and it cannot explain the significance in the view of generalization ability. The existing performance evaluation method of target tracking is a posteriori assessment method. The common target tracking performance evaluation method is the root mean square error (RMSE), which is the root mean square error between the estimated state and the true state of the target in a specific scene. Its advantages are intuitive and high confidence. However, RMSE is a method of posterior performance evaluation method with blindness. If there is a prior performance evaluation method that can predict the comprehensive performance and subdivided performance evaluation value of the target tracking method in advance, it can guide selecting and improving the target tracking method in advance, which can effectively overcome the blindness. For the last issue, improve the performance of underwater multi-target tracking. The original intention of any target tracking method is to obtain the target’s real-time state quickly and accurately. As for the special underwater environment, firstly, we should be clearly aware the advantages and disadvantages of the current target tracking method and the underwater environment, that’s to say, the comprehensive performance evaluation and the subdivided performance evaluation of each influencing factor. Then we grasp, focus and solve the main contradiction, faster and better improving target tracking performance. This links the first two questions with the last questions. As for the specific underwater environment, we propose some new and modified method to improve the target tracking performance for several aspects, such as track initiation method, data association method, track termination method, and underwater high-precision navigation method.

语种中文
产权排序1
文献类型学位论文
条目标识符http://ir.sia.cn/handle/173321/21790
专题海洋信息技术装备中心
作者单位1.中国科学院沈阳自动化研究所
2.中国科学院大学
推荐引用方式
GB/T 7714
李冬冬. 强杂波环境水下多目标跟踪方法关键技术研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2018.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
强杂波环境水下多目标跟踪方法关键技术研究(3666KB)学位论文 开放获取CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[李冬冬]的文章
百度学术
百度学术中相似的文章
[李冬冬]的文章
必应学术
必应学术中相似的文章
[李冬冬]的文章
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。