节点文献

MEMS-SINS/GPS组合导航关键技术研究

Research on the Key Technologies of MEMS-SINS/GPS Integration Navigation Syetem

【作者】 崔留争

【导师】 贾宏光;

【作者基本信息】 中国科学院研究生院(长春光学精密机械与物理研究所) , 机械制造及其自动化, 2014, 博士

【摘要】 基于微机电系统(Micro-Electro Mechanical Systems, MEMS)技术的捷联惯性导航系统(Strapdown Inertial Navigation System, SINS)与全球定位系统(GlobalPositioning System, 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问题为目标,以惯性辅助的紧耦合组合结构为框架,研究了MEMS-惯性测量单元(Inertial Measurement Uunit, IMU)误差分析与补偿、惯性辅助的接收机跟踪环路设计、组合导航滤波器设计等关键技术。主要工作内容可概括为:(1)为建立MEMS-IMU高精度误差模型,研究了其误差特性。建立了确定性误差的系统误差模型,设计了标定与补偿方案,提高了使用精度;采用Allan方差分析法辨识了随机误差的主要误差项与误差参数,建立了随机微分方程模型。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了设计输入。(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结构。建立了SINS的非线性误差模型,推导了伪距差与伪距率差的非线性量测模型,为组合导航滤波器设计提供了依据;设计了惯性辅助的接收机跟踪环路,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卫星信号失锁问题,且具有较强的工程可实现性。(3)针对组合导航模型中的非线性问题,分析了扩展卡尔曼滤波(ExtendedKalman Filtering, EKF)与无迹卡尔曼滤波(Unscented Kalman Filter, UKF)的滤波精度与计算复杂度,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,其滤波精度与UKF相当,与EKF相比提高了30%以上,执行时间与UKF相比降低了45%。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,提出了径向基函数神经网路(Radial Basis Function Neural Network, RBFNN)辅助自适应卡尔曼滤波的信息融合方法,设计了RBFNN在线训练方法与卡尔曼滤波的自适应算法;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。该方法可在GPS失锁时有效阻尼导航误差发散。(5)研制了MEMS-SINS/GPS惯性辅助紧耦合组合导航系统原理样机,建立了惯性器件标定与测试系统、组合导航半物理仿真试验系统。通过半物理仿真试验、跑车试验、高动态试验等,验证了相关技术,测试了主要性能指标,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°。

【Abstract】 The integration of Strapdown Inertial Navigation System (SINS) based on thetechnology of Micro-Electro Mechanical Systems (MEMS) and Global PositioningSystem (GPS) provides a feasible solution for the implementation of navigationsystem featured with low-cost, small scale and lightweight. With the implementationof the navigation system of UAV as background, the inertial-assisted tightly-coupledstructure as a framework, this dissertation studies the key technologies, whichinvolves MEMS-Inertial Measurement Uunit (IMU) error analysis and compensationscheme, inertial-assisted receiver tracking loop design, integrated navigation systemfilter design, to improve the accuracy and the reliability of MEMS-SINS/GPSintegrated navigation system.The main accomplishments are listed below:(1) The error analysis and compensation scheme has been studied to establishthe high-precision error model for MEMS-IMU. As to the systematic error, the modeland calibration schemes are proposed for error mitigation; as to the stochastic error,the major error terms and parameters are identified using Allan variance for errormodeling. The error model provides design input for the simulation of inertialinstrument and the design of integrated navigation system filter.(2) The inertial-aided tightly-coupled structure is designed to solve the problemof the GPS outages in high dynamic environment. The nonlinear model for SINS error is analyzed and established, and the nonlinear measurement model for thepseudo-range difference and pseudo-range rate difference is derived, providing thebasis for the filter design. An approach based on the concept of feedforward toreconfigure the PLL model is introduced, eliminating the effects of dynamic stress.High dynamic experiment shows the reliable tracking for the signal is possible underthe dynamic circumstances of50g/s. The structure solves the problem of loss of lockcaused by the dynamics stress fundamentally and practicality.(3) As to the nonlinearity issue in integrated navigation model, the filteringaccuracy and calculation complex of Extended Kalman Filtering (EKF) andUnscented Kalman Filter (UKF) are studied, the improved UKF filtering algorithm isproposed, with UKF executing time update and EKF executing sequencialmeasurement update. Simulation result demonstrates that the accuracy of theimproved algorithm is the same as UKF, but30%better than EKF, and the calculationtime has been reduced by45%compared with UKF, meeting the system accuracy andreal-time requirements.(4) To make MEMS-based SINS/GPS meet the accuracy requirements duringGPS outages, Radial Basis Function Neural Network (RBFNN) aided adaptiveKalman filtering information fusion method is proposed. RBFNN training strategyand Kalman filtering measurement noise adaptive algorithm are designed. Vehicleexperiment shows that the position error is within15m during40s GPS outages; andwithin90m during100s GPS outages. The proposed method can effectively damp thedivergence of the navigation error during the GPS outages.(5) The prototype of the MEMS-SINS/GPS inertial-assisted tightly-coupledintegrated navigation system is designed; the calibration system for inertial sensor andthe hardware-in-loop simulation system are established. The proposed system solutionhas been validated by means of hardware-in-loop simulation, dynamic test, andvehicle experiment, result shows that position error is within7m, velocity error within0.4m/s, attitude error within0.2°and bearing error within0.6°.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络