MEMS-SINS/GPS组合导航关键技术研究

	MEMS-SINS/GPS组合导航关键技术研究
	崔留争
学位类型	博士
导师	贾宏光
	2014-07
学位授予单位	中国科学院大学
学位专业	机械制造及其自动化
摘要	MEMS-SINS与GPS的组合为实现导航系统的低成本、小型化与轻量化设计提供了可行方案。本文以小型无人飞行器导航系统的研制为背景，以解决MEMS-SINS/GPS存在的精度低与可靠性差的问题为目标，以惯性辅助的紧耦合组合结构为框架，对MEMS-IMU误差分析与补偿、惯性辅助的接收机跟踪环路设计、组合导航系统建模与滤波器设计等关键技术进行了研究，并对系统进行了实现与试验验证。主要工作内容可概括为：　　（1）为建立MEMS-IMU的高精度误差模型，研究了其误差特性。对于确定性误差，建立了系统误差模型，设计了标定与补偿方案，提高了使用精度；对于随机误差，利用Allan方差分析方法辨识了陀螺与加速度计的主要误差项与误差参数，建立了随机误差模型。MEMS-IMU误差模型的建立为惯性器件仿真与组合导航滤波器设计提供设计输入。　　（2）针对高动态环境下GPS接收机信号失锁问题，设计了惯性辅助的紧耦合组合导航结构方案。分析并建立了SINS非线性误差模型，推导了伪距差与伪距率差的非线性量测模型，为滤波器的设计提供依据；建立了惯性辅助的接收机跟踪环路模型，通过引入前馈通道，消除了载体动态应力的影响，试验表明，在50g/s的动态条件下，可保证对信号的可靠跟踪。该组合导航结构从根本上解决了载体动态导致的信号失锁问题，同时具有较强的工程可实现性与可操作性。　　（3）针对惯性辅助的紧耦合组合导航模型中的非线性问题，分析了EKF与UKF的滤波精度与计算复杂，提出了改进UKF算法，由UKF进行时间更新，EKF进行序贯量测更新；仿真结果表明，滤波精度与UKF相当，与EKF相比提高了30%以上，执行时间与UKF相比降低了47%，改进UKF算法可同时满足系统对精度与实时性的要求。　　（4）针对GPS信号失锁时的滤波发散问题，提出了RBFNN辅助自适应Kalman滤波的组合导航滤波器结构，设计了RBFNN在线训练方法与Kalman滤波的自适应算法；跑车试验结果表明，在GPS信号断开时间为40s时，位置误差优于15m；断开时间为100s时，位置误差优于90m，该方法能在GPS失锁时对导航误差发散进行有效阻尼。（5）在系统实现与试验验证方面，设计了MEMS-SINS/GPS惯性辅助紧耦合组合导航系统原理样机，分别建立了惯性器件标定与测试系统、组合导航半物理仿真系统。通过半物理仿真实验、跑车试验、高动态实验对系统的关键技术进行了验证与测试，结果为：位置精度优于7m，速度精度优于0.4m/s，水平姿态精度优于0.2°、航向精度优于0.6°。
其他摘要	The integration of MEMS-SINS and GPS provides a feasible solution for the implementation of navigation system featured with low-cost, small scale and lightweight. With the implementation of the navigation system of UAV as background, the inertial-assisted tightly-coupled structure as a framework, this paper studies the key technology, which involves the MEMS-IMU error analysis and compensation scheme, inertial-assisted receiver tracking loop design, integrated navigation system modeling and filter design, to improve the accuracy and the reliability of MEMS-SINS/GPS. The main accomplishments are listed below:　　(1) The error analysis and compensation scheme has been studied to establish the high-precision error model for MEMS-IMU. As to the systematic error, the model and calibration schemes are proposed for error mitigation; as to the stochastic error, the major error terms and parameters in the gyro and accelerometer output are identified using Allan variance for error modeling. The error model provides design input for the simulation of inertial instrument and the design of integrated navigation system filter. 　　(2) The inertial-aided tightly-coupled structure is designed to solve the problem of the GPS outages in high dynamic environment. The nonlinear model for SINS error is analyzed and established, and the nonlinear measurement model for the pseudo-range difference and pseudo-range rate difference is derived, providing the basis for the filter design. An approach based on the concept of feedforward to reconfigure the PLL model is introduced, eliminating the effects of dynamic stress. High dynamic tests shows the reliable tracking for the signal is possible under the dynamic circumstances of 50g/s. The structure solves the problem of loss of lock caused by the dynamics stress fundamentally and practicality.　　(3) As to the nonlinearity issue in integrated navigation model, the filtering accuracy and calculation complex of EKF and UKF are studied, the improved UKF filtering algorithm is proposed, with UKF executing time update and EKF executing sequencial measurement update. Simulation result demonstrates that the accuracy of the improved algorithm is same as UKF, 30% better than EKF, and the calculation time has been reduced by 47% compared with UKF, meeting the system accuracy and real-time requirements.　　(4) To make MEMS-based SINS/GPS meet the accuracy requirements during GPS outages, radial basis function neural network (RBFNN) aided adaptive Kalman filtering information fusion method is proposed. RBFNN training strategy and Kalman filtering measurement noise adaptive algorithm are designed. Vehicle experiment shows that the position error is within 15m during 40s GPS outages; and within 90m during 100s GPS outages. The proposed method can effectively damp the divergence of the navigation error during the GPS outages.(5) For the systematic implementation and experimental verification, the prototype of the MEMS-SINS/GPS inertial-assisted tightly-coupled integrated navigation system is designed; the calibration and validation system for inertial sensor and hardware-in-loop simulation system are established. The proposed system solution has been validated by means of hardware-in-loop simulation, dynamic test, and vehicle experiment, result shows that position error is within 7m, velocity error within 0.4m/s, vertical attitude error within 0.2° and bearing error within 0.6°.
语种	中文
文献类型	学位论文
条目标识符	http://ir.ciomp.ac.cn/handle/181722/41398
专题	中科院长春光机所知识产出
推荐引用方式 GB/T 7714	崔留争. MEMS-SINS/GPS组合导航关键技术研究[D]. 中国科学院大学,2014.

条目包含的文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
崔留争.pdf（1531KB）			开放获取	CC BY-NC-SA	请求全文