【作者】 陈益；

【导师】 薄煜明；

【作者基本信息】 南京理工大学 ， 控制科学与工程， 2010， 博士

【摘要】 光电测量装置是现代化战争中获取信息的重要手段。各种装载于运动载体上的光电设备,必须考虑探测器视轴对基座运动的隔离,以及在此基础上对被测目标的精密跟踪、激光的高精度照准或是光电对抗中光束的精确定向等要求,具有上述功能的光电设备被称为稳定跟踪系统。本文通过引入“目的域”这一全新概念,将稳定跟踪的控制过程区分为跟踪与丢失两类事件,并以此作为研究出发点,深入分析其中涌现出的理论与实践问题,完成的主要工作包括：1.运用刚体机构学原理,分别从角位置补偿和角速度补偿的角度推导了实现视轴稳定的机理,分析了自动跟踪状态下系统的工作原理,并给出了两轴环架结构中存在的一般问题与解决方法；在考虑环架结构间耦合因素的基础上,建立了执行稳定跟踪任务的机电系统各环节的数学模型；概述了光电稳定跟踪系统的误差及其主要来源,通过结构框图和传递函数分析了各误差源的作用途径,并借助噪声白化理论,构建了稳定跟踪控制问题的一般理论框架。2.将稳定跟踪误差分析由空间域延展为时空域,以均值、均方差、自然频率为基本元素给出了误差序列对指定目的域的穿越特性,进而提炼出跟踪中断概率这一稳定跟踪系统的关键性指标,并将其转化为一组由滞留时间、超差时间以及随机穿越周期等构成的概率特征量指标集约束。3.直接从工程实际需求出发,提出以区域极点形式表征的快速性指标、动态误差系数或品质因数形式表征的准确度指标、稳态方差形式表征的精确度指标,扰动抑制H∞范数形式表征的鲁棒性指标以及概率特征量和统计特征量形式表征的随机穿越特征指标描述稳定跟踪系统的性能要求；分别基于状态反馈控制和PID控制,推导出特定目的域下随机穿越特征量指标的解析表达,分析了其与不同性能指标间的相容性；当预先选定的期望指标集相容时,提出了一种利用双群体机制下的差分进化算法和线性矩阵不等式方法相结合对控制器参数进行多指标满意优化求解的策略。4.针对某分布式火控系统中光电观瞄子系统的实际情况,探讨了利用光电设备对基座姿态信息进行视觉估计测量的方法,以地面上共面设置的四个固定合作信标和载体平台上双焦CCD摄像机组成单目视觉系统,分析了利用静态解析算法对姿态角进行解算时的可行性与不足；提出采用刚体运动学原理与二阶时间相关模型建立姿态参数的动态估计系统,并引入渐消记忆法改进后的平方根UKF滤波算法,用于抑制由于系统建模误差所导致的估计精度下降。5.搭建了基于嵌入式微处理器的观瞄子系统实验平台,简要介绍了伺服控制器的主要硬件模块和软件设计流程；依据不同的控制任务要求,设计了多种性能指标约束下的分段控制策略,并通过物理转台仿真和外场跟踪实验,进一步总结验证了全文的主要理论成果。更多还原

【Abstract】 Photoelectric measuring instruments are the great tools for gaining information in modern warfare. Various kinds of photoelectric devices mounted on mobile vehicles have to take account of the isolation for line-of-sight (LOS) from vehicles’motion, so as to accomplish the precision target tracking, laser aiming, or beam orienting in particular cases. Equipments integrated the above functions are known as the stabilized tracking systems. In this dissertation, a brand-new concept of "target area" was introduced, and the stabilized tracking process was thus divided into two events:target tracking and target losing. Based on this principal, the relevant theoretical and practical issues were deeply analyzed, and the primary work was organized as follows:1. Using rigid body kinematics, the isolating principium of LOS against base turbulence was illuminated from the aspects of angular velocity compensation and angle position compensation respectively, and the auto tracking principle as well as the general considerations of two-axis gimbals was also given. Based on inertia coupling factors, the mathematics models of electro-mechanical system which performs the stabilization and tracking tasks were discussed in detail. The errors and their main origins in photoelectric stabilized tracking systems were outlined, and the structure chart together with transfer function manner were used to analyze the mechanism of these error sources. Following this, the theoretical framework of stabilized tracking control problems with multiple indices constraints was constructed by aid of noise whitening filter.2. The relevant concepts of target area, residence state, exceeding tolerance state, and random passage period frequency were clarified, and on this basis the stabilization and tracking error analysis were extended from conventional spatial domain to time-space domain. The passage characteristics on a certain target area were provided by taken mean value, mean-square deviation and natural frequency as basic elements, and then the tracking interruption probability, a key index of the system was put forward, which was further transformed to a set of probability distribution characteristic indices combined by residence time, exceeding tolerance time, and random passage period time.3. The desired performance indices of stabilized tracking systems were concludes as rapidity index symbolized by regional poles, accuracy index symbolized by dynamic error coefficients or quality factors, precision index symbolized by steady error covariance, robust index symbolized by disturbances rejection H-infinite norm, and the stochastic passage indices symbolized by probability distribution and statistics characteristics. Based on state feedback control and PID control strategy, the stochastic passage characteristic (SPC) indices on zonal target area were deduced respectively, and the consistency between different performance indices was analyzed. When the above indices are consistent, a modified differential evolution (DE) algorithm was utilized together with linear matrix inequality (LMI) method to find out a set of satisfactory optimal parameters for the certain control strategy.4. In light of the actual situation of a photoelectric observation/aiming subsystem in a distributed fire control system, a method for estimating the attitude of aerostat was proposed. A monocular vision system was established by four coplanar reference points on the ground and a double-focus CCD camera in the carrier, and the feasibility and shortage of an analytical algorithm were analyzed. Then a second-order time-correlation model was applied to obtain the state equations of the system, and the fading memory factor was introduced to reform the square-root unscented Kalman Filter (SR-UKF), which was able to restrain the estimating accuracy decline caused by system modeling error.5. Experimental equipments were set up for the photoelectric observation/aiming subsystem based on embedded microprocessor unit, and the hardware modules as well as the software flow chart of the controller design was briefly described. According to the different task requirements, a piecewise control strategy with multi-performance indices constraints was devised. The final physical simulation and field tracking test have validated the chief theoretical achievements of this dissertation.更多还原