自主导航与完好性的研究

发布时间：2018-06-21 09:39

本文选题：惯性导航系统 + 卡尔曼滤波　；参考：《电子科技大学》2014年硕士论文

【摘要】：随着技术的进步,原本应用于军事领域的各种导航技术已应用于民用领域或正在向民用领域推广。在民用领域,导航系统的安全性至关重要,完好性作为对导航系统安全性的一种衡量是评价一个导航系统优劣的重要标准。自主导航是一种不借助人工设置的目标和信息源的导航仪器来引导物体运行的导航方法,惯性导航就是一种自主导航,包含惯性导航的组合导航也可以归于自主导航。导航系统需要获得测量信息进行导航解算,本文将重点研究如何通过故障检测来提高导航系统的完好性。首先,对惯性导航系统进行仿真分析。惯性导航系统为了确定运动物体的初始姿态要先进行初始对准。初始对准包括粗对准和精对准,精对准用卡尔曼滤波器估计粗对准的误差。本文提出了改进的精对准算法,利用两个加速度计的输出和陀螺仪的三个输出共五个值作为卡尔曼滤波器的观测值。建立数学模型模拟陀螺仪和加速度计的输出,并搭建了惯性导航系统仿真平台,对改进的精对准进行了仿真分析。仿真结果显示改进的精对准算法提高了对方位误差角的估计精度。其次,仿真分析了GPS导航系统,并在惯性导航系统平台上结合GPS导航系统,分别仿真松组合导航和紧组合导航,对不同导航系统的仿真结果进行比较分析。搭建的仿真平台为之后的故障检测研究做铺垫。然后,对故障检测展开研究。首先介绍最优奇偶矢量法,它是多故障检测的基础,发现最优奇偶矢量法存在缺陷,不同观测值中的故障有可能使残差变小,不利于对故障的检测。为了解决这个问题,本文提出了改进的最优奇偶矢量法,仿真比较传统的方法和改进的方法对故障的检测效果,仿真结果表明改进的最优奇偶矢量法有效地解决了残差变小的问题。实际情况中观测精度可能不同,仿真比较残差标准化法和加权奇偶矢量法对故障的检测效果,分析加权奇偶矢量法的优势。另外本文发现,当卡尔曼滤波器中待估值的个数远远大于观测值的个数的时候,无法使用最优奇偶矢量法检测观测方程中的故障,因此构造新的统计量检测其中的故障,仿真验证了新的统计量对故障具有较好的检测效果。最后,研究在GPS和组合导航系统中引入故障检测的方法和可行性。仿真验证了本文提出的故障检测算法在这些导航系统中能有效地检测出故障。
[Abstract]:With the development of technology, all kinds of navigation technology used in military field have been applied to civilian field or are being popularized to civilian field. In civil field, the security of navigation system is very important. As a measure of navigation system security, completeness is an important criterion to evaluate a navigation system. Autonomous navigation is a kind of navigation method which can guide the operation of objects without the aid of the navigation instrument of the target and information source. Inertial navigation is a kind of autonomous navigation, and the integrated navigation including inertial navigation can also be attributed to autonomous navigation. Navigation system needs to obtain measurement information for navigation solution. This paper will focus on how to improve the integrity of navigation system through fault detection. First, the inertial navigation system is simulated and analyzed. In order to determine the initial attitude of the moving object, the inertial navigation system needs initial alignment. The initial alignment consists of coarse alignment and fine alignment. Kalman filter is used to estimate the error of coarse alignment. In this paper, an improved precision alignment algorithm is proposed, which uses the output of two accelerometers and three outputs of gyroscopes as the observed values of Kalman filter. The mathematical model is established to simulate the output of gyroscope and accelerometer, and the simulation platform of inertial navigation system is built. The improved precision alignment is simulated and analyzed. Simulation results show that the improved precision alignment algorithm improves the precision of azimuth error angle estimation. Secondly, the GPS navigation system is simulated and analyzed, and the loose integrated navigation and compact integrated navigation are simulated on the inertial navigation system platform. The simulation results of different navigation systems are compared and analyzed. The simulation platform is used to pave the way for the later fault detection research. Then, the research on fault detection is carried out. This paper first introduces the optimal even-odd vector method, which is the basis of multi-fault detection. It is found that the optimal odd-even vector method has defects, and the fault in different observation values may make the residual error smaller, which is not conducive to fault detection. In order to solve this problem, an improved optimal even-odd vector method is proposed in this paper. The simulation results are compared with the traditional method and the improved method for fault detection. The simulation results show that the improved optimal even-odd vector method can effectively solve the problem of reducing the residual error. The actual observation accuracy may be different. The residual standardization method and the weighted odd-even vector method are compared in the fault detection effect, and the advantages of the weighted odd-even vector method are analyzed. In addition, it is found that when the number of Kalman filters to be estimated is far greater than the number of observed values, it is impossible to detect the faults in the observation equation by using the optimal even-odd vector method, so a new statistic is constructed to detect the faults. The simulation results show that the new statistic is effective for fault detection. Finally, the method and feasibility of introducing fault detection into GPS and integrated navigation system are studied. Simulation results show that the proposed fault detection algorithm can effectively detect faults in these navigation systems.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN96

【参考文献】