【作者】 王海龙；

【导师】 刘雨；

【作者基本信息】 哈尔滨工业大学 ， 控制科学与工程， 2015， 硕士

【摘要】 系统级的标定技术与单个仪表在实验室中的测试很不一样。仪表的一些误差系数安装到系统中后,往往会随着环境条件改变而发生改变,需要再重新标定。基于惯性稳定平台的惯导平台系统包含了具有三个旋转自由度的框架系统,这种旋转功能使得它能够不必依赖于外部测试设备或基准实现自标定。连续翻滚测试方法与当前广泛使用的多位置翻滚测试方法相比,由于平台一直处于伺服工作状态,不仅能够充分利用翻滚过程中全部的观测信息,辨识出更多的误差项系数,而且有更高的标定精度,测试过程也简单高效。但是误差模型方程和试验设计过程也很复杂。本文对惯导平台系统连续翻滚自标定试验设计和辨识相关问题展开研究,首先选用加速度计和陀螺仪共计30个误差项系数,状态方程和观测方程分别使用ψ角和加速度计输出建立。为了提高系统误差参数的可观测度,得到连续翻滚试验中平台的最优旋转轨迹,使用D最优化试验设计方法,得到相应的数学表达式。通过适当的数学描述和工程简化,将最优连续旋转轨迹的设计问题转化为最优控制问题进行求解。针对最优试验设计的求解问题,采用全局智能优化算法求解。同时为了提高最优轨迹的计算效率和精度,将非线性约束最优化问题转换为无约束最优化问题求解,引入了壁垒函数和改进的RSSA算法。新方法的求解性能优于传统的遗传算法。仿真结果表明,算法通过合理的参数配置,不仅极大地提高了D最优设计求解的计算效率,并且得到的最优轨迹适应度值精度要好于传统的遗传算法。在平台连续旋转最优轨迹设计结果的基础上,对连续翻滚自标定试验的误差辨识方法进行研究。在之前建立的基于ψ的系统误差模型基础上,引入余弦变换矩阵,得到加速度计测量误差的垂直分量作为新息对加速度计进行辨识,加速度误差的水平分量作为新息对陀螺仪进行辨识,提出了双卡尔曼滤波的辨识方法,将陀螺仪和加速度计通过解耦分开辨识,通过仿真结果验证了双卡尔曼滤波辨识方法的有效性。更多还原

【Abstract】 The calibration technology of system level is far different from that of a single instrument in the laboratory test. Some error coefficients of instruments tend to change as environmental conditions change, which need to be recalibrated. Inertial navigation platform system(INPS), which is based on inertial stabilized platform, contains a system of framework with three rotational degrees of freedom. The rotational function enables it to realize self-calibration not rely on external reference test equipment or reference. Continuous tumbling test method is compared with the multi-position testing method which is used broadly now, because the platform works always in servo condition, not only can make full use of all the observing information during the tumbling process and identify more error coefficients, but also has a higher calibrating accuracy and the testing process is simpler and more efficient. On the other hand, the error model equation and experimental design process for continuous tumbling test is more complicated.The research in this thesis is about the related problems in continuous tumbling self-calibration experimental design and identification methods of error model parameters for inertial navigation platform system. First, a total of 30 error model parameters of the three accelerometers and three gyroscopes are selected, state equations and observation equations are respectively established using Y angles and accelerometer outputs. In order to improve the observability degree of parameters to be identified and obtain optimal rotation trajectory of platform continuous tumbling experiment, using D optimal experimental design method to get the corresponding mathematical expression. By means of appropriate engineering simplification and mathematical deduction, the optimal design problem of continuous tumbling test is transformed into an optimal control problem to be solved.According to the solution of the optimal design of experiment, using the global intelligent optimization algorithm. At the same time in order to improve the computational efficiency and the precision of the optimal trajectory, Barrier function and the improved RSSA algorithm are introduced, which transforms the nonlinear constrained optimization problem into unconstrained optimization problem. The computing performance of new method is superior to the traditional genetic algorithm. The simulation results show that reasonable configuration for the algorithm parameters, not only greatly improves the efficiency of solving the D optimal design problem, but also the fitness values of optimal trajectory obtained is better than traditional genetic algorithm.Based on the results of the optimal trajectory design of the platform continuous tumbling experiment. The error identification method for continuous tumbling self-calibration test in the thesis is studied. On the basis of the established system model based on Y angles, we introduce the cosine transformational matrix. The innovation sequence for the accelerometer calibration is the vertical component of the acceleration error, and the innovation sequence for the gyro calibration is the horizontal component of the acceleration error. The identification method of dual kalman filter is proposed to separate gyroscope and accelerometer identification by decoupling them, we can verify the effectiveness of the double-kalman-filter identification method by the simulation results.更多还原