CASIA OpenIR  > 毕业生  > 硕士学位论文
典型机动目标跟踪与精确预测技术
其他题名Typical Maneuvering Target Tracking and Precise Prediction Technology
张峰
学位类型工程硕士
导师薛文芳
2012-05-24
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业控制工程
关键词输入估计 贝叶斯滤波 交互式多模型算法 自适应转弯率 渐消因子 Input Estimation Bayesian Filtering Interactive Multiple Model Adaptive Turn Rate Fading Factor
摘要机动目标跟踪技术在现代军事和民用领域均占有非常重要的地位。随着现代航空航天技术的飞速发展,各种飞行器的航行速度和机动性越来越高。机动呈现出来的复杂性、随机性和多样性等特点使得对机动目标的跟踪无论在理论上还是在实践上都较为困难。虽然目前已有不少针对目标跟踪领域的研究,但总的来说,专门针对高速、大机动目标以及低速、灵活机动目标的研究还不能满足工程上的需要。 论文首先深入分析了机动目标跟踪的理论基础,重点对贝叶斯滤波算法进行了归纳总结,并将现有的滤波方法在贝叶斯框架下进行了分类。然后,论文将典型的机动目标跟踪分为两类:对高速、大机动目标的跟踪和对低速、灵活机动目标的跟踪,并分别针对其运动特征提出了基于渐消因子的修正输入估计算法和一种自适应转弯率的优化方法,然后本文结合仿真实例对文中提出的两种改进算法分别进行了仿真实验,实验结果验证了改进算法的有效性。论文取得的主要成果如下: (1)贝叶斯滤波算法的分析 论文从贝叶斯理论出发,对其原理进行了详细的分析,将滤波算法归纳为最优滤波算法和次优滤波算法,并分别对其进行了深入的研究分析,找出了现有算法的优缺点。 (2)论文首先在已有输入估计算法的基础上引入了渐消因子,提出了基于渐消因子的修正输入估计算法。改进算法较现有的输入估计算法的优势在于:不需要机动检测环节,在跟踪大机动目标时能够在一定程度上减弱状态突变引起的跟踪性能恶化等问题。然后,论文将改进算法应用在高速、大机动目标的跟踪中,仿真实验结果表明,改进算法较现有算法跟踪性能有明显的提高。 (3)论文在现有自适应转弯模型的基础上优化了自适应转弯率的计算方法,改进算法较现有算法的优势在于:能够更加准确地估计出转弯率的大小和方向;因为状态转移矩阵和系统噪声协方差矩阵的关键参数是转弯率,因而用改进算法计算出来的状态转移矩阵和系统噪声协方差矩阵也更加准确,因此,改进算法对转弯运动机动目标的跟踪精度也更高。最后,论文将改进算法应用在低速、灵活机动目标的跟踪中,仿真实验结果表明,改进算法较现有算法跟踪性能有明显的提高。
其他摘要In both military and civilian fields, maneuvering target tracking technology plays an important part. With the rapid development of modern aeronautics and astronautics technology, flight vehicles get faster and faster, and their maneuverability gets higher and higher as well. The characteristics of complexity, randomness and multiplicity resulted from maneuver further make the tracking task very difficult both theoretically and practically. Although lots of research work has been established on maneuvering target tracking, generally speaking, research aiming at tracking of both high-speed & large-maneuver and low-speed & flexible-maneuver targets is still not enough for engineering requirement. First, we analyze the basic theory of maneuvering target tracking deeply, of which Bayesian filtering algorithms are mainly summarized, and existing filtering methods are classified in the framework of Bayesian theory. Second, typical maneuvering target tracking problems are classified into two kinds: high-speed & large-maneuver target tracking and low-speed & smart-maneuver target tracking, and modified input estimation algorithm based on fading factor and an improved adaptive turn rate calculating method are proposed respectively according to their motion characteristics. Third, new algorithms are simulated respectively by simulation instances, and simulation results verify the validity of the new algorithms. The main fruit of the dissertation is as follows: (1)Analysis on Bayesian filtering algorithm We start from the Bayesian theory, and make an exhaustive analysis on its principle. Existing filtering algorithms are classified to optimal filtering algorithms and suboptimal filtering algorithms. And then we make a deep study and find out the merits and drawbacks of the existing algorithms. (2) We introduce a fading factor into the existing input estimation algorithm and propose a new algorithm called modified input estimation algorithm based on fading factor. The strengths of the new algorithm compared with the existing ones are as follows: without maneuver detection, able to resolve problems like tracking performance degrade resulted from state mutation. Then we apply the new method to the tracking of high-speed & large-maneuver targets. Simulation results show that the new method has a better tacking performance than the existing methods. (3) We optimize the adaptive turn rate calculating method based on existing methods. The strengths o...
馆藏号XWLW1802
其他标识符2009M8014628006
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/7622
专题毕业生_硕士学位论文
推荐引用方式
GB/T 7714
张峰. 典型机动目标跟踪与精确预测技术[D]. 中国科学院自动化研究所. 中国科学院研究生院,2012.
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
CASIA_2009M801462800(1948KB) 暂不开放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
[张峰]的文章
百度学术
百度学术中相似的文章
[张峰]的文章
必应学术
必应学术中相似的文章
[张峰]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。