【作者】 王洋；

【导师】 王庆超；

【作者基本信息】 哈尔滨工业大学 ， 一般力学与力学基础， 2008， 硕士

【摘要】 目标跟踪无论在空中侦察与预警、弹道导弹防御、战场监视等军事领域,还是空中交通管制、交通导航等民用领域均有广泛的应用,扮演着重要的角色。然而随着现今各类飞行器机动性能的提高,传统的目标跟踪算法已逐渐不能满足精确跟踪的要求,应用自适应滤波提高机动目标跟踪的性能,是一项颇具现实意义的研究课题。机动目标跟踪自适应滤波算法主要分三种类型,检测自适应滤波、实时辨识自适应滤波与全面自适应滤波,本文首先对三种算法中各自典型算法:变维算法(VDF)、当前统计模型均值自适应算法、交互式多模型算法(IMM)算法进行研究,通过仿真分析出综合性能上IMM算法有着明显的优势,而当前统计模型均值算法在高机动的情况下跟踪性能良好。当前统计模型算法存在着两个缺陷:弱机动目标跟踪能力差;目标实现跟踪需要对目标机动加速度进行合理的先验假设。本文深入研究当前统计自适应算法原理,应用模糊机理对当前统计自适应算法中先验假设进行适当修正,仿真表明其在一定程度上弱化了当前统计模型的缺陷。针对IMM算法中模型转移概率矩阵为固定值的情况,应用滤波过程中得到的模型概率,实时对模型状态转移矩阵进行修正,并对基于常速(CV)、常加速(CA)、机动转弯模型(CT)的改进IMM算法与传统IMM算法进行了对比仿真,验证了改进算法具备较高的精度,机动调节时间也相应减少。最后考虑到当前统计模型对高机动目标的良好跟踪能力,将改进当前统计模型融入修正转移概率的IMM算法中,结合用以描述非机动、弱机动的CV、CA模型作为IMM算法子模型集,增强了IMM算法的自适应性。更多还原

【Abstract】 Maneuvering Target Tracking is particularly important in the field of national defense research. In the air reconnaissance and warning, missile defense, battlefield surveillance and other military fields, as well as air traffic control, navigation and other civilian traffic areas, Maneuvering Target Track is a key technology. Now as the mobility of modern aircrafts increasing, the traditional methods of target tracking could not satisfy the request of tracking accurate at all. So using the adaptive filtering algorithms to improve the performance of Maneuvering Target Tracking is a meaningful research work.The basic adaptive filtering algorithms of Maneuvering Target Tracking can divide into three parts: Examines adaptive filtering, Real-time identification adaptive filtering and Comprehensive adaptive filtering. In this paper we research on Variable Dimension Filtering algorithm, Current Statistical Model Adaptive algorithm and Interactive Multiple Model (IMM) which belongs to the basic algorithms respectively. Through the simulation analysis, IMM algorithm has the obvious superiority in the comprehensive performance, and Current Statistical Model algorithm in the case of large maneuvering can get good tracking performance.Current Statistical Model algorithm has two flaws. One is the algorithm isn’t suitable for slight maneuvering, and the other is algorithm needs reasonable transcendental hypothesis for the acceleration of targets. Based on the study of Current Statistical Model algorithm, an improved Current Statistical Model algorithm using fuzzy theory is proposed to revise the two flaws reasonably. And the algorithm is proved by computer simulation.In the traditional IMM algorithm the model probability transfer matrix is a fixture. Here we proposed a new algorithm. This algorithm using the model probability, which obtains from the process of IMM algorithm, revises the model probability transfer matrix on time. For comparing with the traditional algorithm we set an IMM algorithm with CV, CA, CT three types of model which is used to simulate. From the results of simulation, we can obtain the new method can get high accurate and less adjust time. At last considering the good tracking ability of the improved Current Statistical Model algorithm, we used the improved Current Statistical Model as a sub-model in new IMM algorithm to describe the situation of target maneuvering so that to expand the adaptive performance of IMM algorithm.更多还原