【作者】 于飞；

【导师】 孙枫；

【作者基本信息】 哈尔滨工程大学 ， 控制理论与控制工程， 2003， 硕士

【摘要】 初始对准是惯性导航系统的关键技术之一，它直接影响惯导系统的导航性能，由于系统中存在有各种随机因素，初始对准的精度和对准时间之间存在着矛盾，对准精度越高所需时间也越长。通过对SINS系统的误差传播特性分析可以发现，系统的误差主要由传感器误差和初始对准误差决定，在对传感器误差进行补偿之后，整个系统的精度几乎就保持在初始对准的精度上；另外，在初始对准过程中还包含陀螺和加速度计的测漂和定标工作，以便进行误差补偿。 初始对准从控制上讲就是施加一定的控制角速度把数学平台转向期望的位置(与地理坐标系重合)，它是按误差方程进行控制的，过去在平台系统中采用频率特性方法，目前已经采用了现代控制理论设计方法。目前尔曼滤波技术的应用为SINS系统的快速对准提供了一种有效的方法，采用卡尔曼滤波进行地面自对准就是对误差角进行最优估计，通过系统校正，使平台坐标系和导航坐标系对准，仿真结果表明卡尔曼滤波应用于SINS系统地面自对准具有很好的结果，然而卡尔曼滤波是在一种理想条件下进行的，即要求系统的动态模型准确并且噪声是统计特性已知的白噪声，在实际应用中这些条件未必能满足，因而卡尔曼滤波的应用还要受到这些前提条件的限制，此外卡尔曼滤波对模型和噪声中的不确定因素也不具备鲁棒性。为解决这一问题，本文将H_∞滤波引入SINS系统的初始对准，设计了具有鲁棒性的H_∞滤波器。仿真结果表明，当有建模误差存在时，H_∞滤波器的鲁棒性要远好于卡尔曼滤波器。更多还原

【Abstract】 Initial alignment is one of the key technology of Inertial Navigation System. It directly influence the navigation performance of navigation system . Because some random factors exist in the system, there are some difference between the precision of Initial alignment and alignment time . Therefore, higher precision, more time. The system error is mainly determined by sensor error and initial alignment error by analyzing the error transfer characteristic of the SINS system. After compensating the sensor error, the precision of the whole system almost remains the Initial alignment process. It should include of gyroscope and speedmeter to compensate the error. Initial alignment is to put on some speed of control angel in order to transferring the maths platform to expected location. It can be controlled by error equation. In the platform system, it used to adopt the modern control theory method.At present , the application of Kalman filtering technology has provided an efficient method for the quick alignment of SINS system. Adopting the Kalman filtering aimed by itself means using it to give the best estimate about the platform error angle. Using the correction system to the correction between the platform coordinate and navigation coordinate. The artificial result shows that the application Kalman filtering for SINS system can bring a good result but Kalman filtering is done in the ideal situation. In other words, it requires the precision of the active model and the noise is the white noise which statistical characteristic can be shown. We could not meet the needs during the actual application, so the application of Kalman filtering has to be limiter by some premise condition. Moreover, Kalmanfiltering does not have robust characteristic to the undetermined factors between model and noise. To solve the problem, the author introduces the H_∞ filtering to the initial alignment and designs the H_∞filtering which owns robust characteristic.更多还原