惯导平台系统自标定实验设计与辨识
发布时间：2019-07-09 11:42
【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。
文内图片：
图片说明：遗传算法仿真图
[Abstract]:The calibration technology at the system level is very different from the test of a single instrument in the laboratory. After some error coefficients of the instrument are installed in the system, they often change with the change of environmental conditions and need to be re-calibrated. The inertial navigation platform system based on inertial stabilization platform contains a frame system with three rotating degrees of freedom, which enables it to realize self-calibration without relying on external test equipment or benchmark. Compared with the multi-position rolling test method, which is widely used at present, the continuous rolling test method is not only able to make full use of all the observed information in the rolling process, identify more error coefficients, but also has higher calibration accuracy and simple and efficient testing process because the platform has been in a servo working state. However, the error model equation and the experimental design process are also very complex. In this paper, the design and identification of continuous roll self-calibration test for inertial navigation platform system are studied. Firstly, a total of 30 error term coefficients of accelerometer and gyroscope are selected, and the equation of state and observation equation are established by using 蠁 angle and accelerometer output, respectively. In order to improve the observable measure of the system error parameters, the optimal rotation trajectory of the platform in the continuous rolling test is obtained, and the corresponding mathematical expression is obtained by using the D optimization test design method. Through proper mathematical description and engineering simplification, the design problem of optimal continuous rotation trajectory is transformed into the optimal control problem. In order to solve the problem of optimal experimental design, the global intelligent optimization algorithm is used to solve the problem. At the same time, in order to improve the computational efficiency and accuracy of the optimal trajectory, the nonlinear constrained optimization problem is transformed into an unconstrained optimization problem, and the barrier function and the improved RSSA algorithm are introduced. The performance of the new method is better than that of the traditional genetic algorithm. The simulation results show that the algorithm not only greatly improves the computational efficiency of D optimal design and solution, but also improves the accuracy of the optimal trajectory fitness value better than the traditional genetic algorithm through reasonable parameter configuration. Based on the design results of the optimal trajectory of continuous rotation of the platform, the error identification method of continuous roll self-calibration test is studied. On the basis of the previous systematic error model based on 蠁, the cosine transform matrix is introduced, and the vertical component of accelerometer measurement error is used as innovation to identify the accelerometer, and the horizontal component of acceleration error is used as innovation to identify the gyroscope. The identification method of double Kalman filter is put forward. The gyroscope and accelerometer are identified separately by decoupling. The effectiveness of the double Kalman filter identification method is verified by the simulation results.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN96
本文编号：2512120
文内图片：
图片说明：遗传算法仿真图
[Abstract]:The calibration technology at the system level is very different from the test of a single instrument in the laboratory. After some error coefficients of the instrument are installed in the system, they often change with the change of environmental conditions and need to be re-calibrated. The inertial navigation platform system based on inertial stabilization platform contains a frame system with three rotating degrees of freedom, which enables it to realize self-calibration without relying on external test equipment or benchmark. Compared with the multi-position rolling test method, which is widely used at present, the continuous rolling test method is not only able to make full use of all the observed information in the rolling process, identify more error coefficients, but also has higher calibration accuracy and simple and efficient testing process because the platform has been in a servo working state. However, the error model equation and the experimental design process are also very complex. In this paper, the design and identification of continuous roll self-calibration test for inertial navigation platform system are studied. Firstly, a total of 30 error term coefficients of accelerometer and gyroscope are selected, and the equation of state and observation equation are established by using 蠁 angle and accelerometer output, respectively. In order to improve the observable measure of the system error parameters, the optimal rotation trajectory of the platform in the continuous rolling test is obtained, and the corresponding mathematical expression is obtained by using the D optimization test design method. Through proper mathematical description and engineering simplification, the design problem of optimal continuous rotation trajectory is transformed into the optimal control problem. In order to solve the problem of optimal experimental design, the global intelligent optimization algorithm is used to solve the problem. At the same time, in order to improve the computational efficiency and accuracy of the optimal trajectory, the nonlinear constrained optimization problem is transformed into an unconstrained optimization problem, and the barrier function and the improved RSSA algorithm are introduced. The performance of the new method is better than that of the traditional genetic algorithm. The simulation results show that the algorithm not only greatly improves the computational efficiency of D optimal design and solution, but also improves the accuracy of the optimal trajectory fitness value better than the traditional genetic algorithm through reasonable parameter configuration. Based on the design results of the optimal trajectory of continuous rotation of the platform, the error identification method of continuous roll self-calibration test is studied. On the basis of the previous systematic error model based on 蠁, the cosine transform matrix is introduced, and the vertical component of accelerometer measurement error is used as innovation to identify the accelerometer, and the horizontal component of acceleration error is used as innovation to identify the gyroscope. The identification method of double Kalman filter is put forward. The gyroscope and accelerometer are identified separately by decoupling. The effectiveness of the double Kalman filter identification method is verified by the simulation results.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN96
【参考文献】
相关期刊论文 前2条
1 于旭东;龙兴武;王宇;张鹏飞;汤建勋;魏国;;激光陀螺单轴旋转惯导系统多位置对准技术研究[J];传感技术学报;2011年06期
2 ;An Adaptive UKF Algorithm for the State and Parameter Estimations of a Mobile Robot[J];自动化学报;2008年01期
相关博士学位论文 前1条
1 杨华波;惯性测量系统误差标定及分离技术研究[D];国防科学技术大学;2008年
，本文编号：2512120
本文链接：https://www.wllwen.com/kejilunwen/wltx/2512120.html
