【作者】 杨帆；

【导师】 刘雨；

【作者基本信息】 哈尔滨工业大学 ， 控制工程（专业学位）， 2019， 硕士

【摘要】 惯性系统已逐渐完成了由平台式向捷联式的过渡,捷联式惯导系统以数字平台取代了实体平台,不仅降低了成本,可靠性也有所提高。但由于惯性测量单元(Inertial Measurement Unit,IMU)被直接安装在机体上,因此系统精度受到了较大影响。本文就是通过提出一种改进的混合优化算法为双轴测试转台设计一个最优旋转轨迹从而提高IMU的动态误差系数的标定精度。IMU角动态误差模型的建立过程中,首先要基于IMU的输出误差建立导航误差方程,然后根据环形激光陀螺IMU和微机械陀螺IMU各自的特点对模型进行简化。然后基于卡尔曼滤波的方法,确定最小化协方差矩阵的迹为初步优化目标。之后又结合转台的特性确定问题中需满足的约束条件,通过罚函数对约束条件进行处理,确定最终优化目标。针对本试验中优化问题,在众多最优化随机搜索算法中,首先选用了不需要导数信息的模式搜索算法和全局搜索性能较好的模拟退火算法这两种算法,分别对其原理进行研究,而后使用两种传统算法对优化目标进行求解,基于优化结果比较两种算法的性能并从三个方面提出对算法的改进方案,最终得到融合了两种算法的混合优化算法。分别采用传统模拟退火算法和改进后的混合优化算法求解本试验中优化目标,进行仿真并绘制各项误差系数标定精度曲线,比较两种算法优化的情况下各项误差系数估计值的大小。从误差精度来看,环形激光陀螺IMU的各项动态误差系数标定精度提高了3%~5%,微机械陀螺IMU的非线性误差标定精度提高了24%~40%,其它动态误差系数标定精度提高了2%~19%。而从计算时间来看,混合优化算法相比于改进参数后单一的模拟退火算法时间增加了5.42%~12.86%。考虑到精度的提高效果,认为这些时间的增加是可以接受的。经验证,本文提出的改进后的混合优化算法是一种高性能的搜索算法,在IMU角动态标定试验中可以有效地提高误差标定精度。更多还原

【Abstract】 The inertial system has gradually completed the transition from platform to strapdown.The strapdown inertial navigation system replaces the physical platform with a digital platform,which not only reduces costs but also improves reliability.However,since the Inertial Measurement Unit(IMU)is directly mounted on the body,the accuracy of the system is greatly affected.In this paper,an improved hybrid optimization algorithm is proposed to design an optimal rotation trajectory for the biaxial test turntable to improve the calibration accuracy of the IMU dynamic error coefficient.In the process of establishing the IMU angular dynamic error model,the navigation error equations are first established based on the output error of the IMU,and then the model is simplified according to the characteristics of the ring laser gyro IMU and the micromachined gyro IMU.Then based on the Kalman filter method,it is determined that the trace of the minimized covariance matrix is a preliminary optimization target.Then,based on the characteristics of the turntable,the constraints that need to be met in the problem are determined,and the constraint conditions are processed by the penalty function to determine the final optimization goal.Aiming at the optimization problem in this experiment,among the many optimization random search algorithms,the two methods of pattern search algorithm without derivative information and simulated annealing algorithm with better global search performance are selected,and the principle is studied separately.Two traditional algorithms are used to solve the optimization target.The performance of the two algorithms is compared based on the optimization results.The improved scheme is proposed from three aspects.Finally,a hybrid optimization algorithm combining two algorithms is obtained.The traditional simulated annealing algorithm and the improved hybrid optimization algorithm are used to solve the optimization target in this experiment respectively.Simulate and plot various error coefficient calibration accuracy curves.Compare the estimated values of each error coefficient under the optimization of two algorithms.From the perspective of error accuracy,the calibration accuracy of the dynamic error coefficients of the ring laser gyro IMU is improved by 3%~5%.For the micro-mechanical gyro IMU,the calibration accuracy of the nonlinear error coefficient is improved by 24%~40%,and the calibration accuracy of other dynamic error coefficients is improved by 2%~19%.From the calculation time,compared to the single simulated annealing algorithm with the improved parameters,the time of the hybrid optimization algorithm increased by 5.42%~12.86%.Considering the improvement in accuracy,it is considered that the increase in time is acceptable.It is verified that the improved hybrid optimization algorithm proposed in this paper is a high-performance search algorithm,which can effectively improve the accuracy of error calibration in the IMU angular dynamic calibration experiment.更多还原