【作者】 吴赛成；

【导师】 王省书； 秦石乔；

【作者基本信息】 国防科学技术大学 ， 光学工程， 2012， 博士

【摘要】 随着我国自主研制的激光陀螺精度逐步提高，以激光陀螺为核心惯性元器件的船用高精度姿态测量技术成为了研究热点。本文以船用高精度激光陀螺捷联姿态测量系统为研究对象，研究了制约进一步提高姿态测量精度的关键因素和解决方法，内容包括惯性测量单元（IMU）系统级标定方法研究、重力场异常对姿态测量系统的影响及补偿方法研究、重力场辅助惯性测量研究、动基座对准研究、滤波延迟补偿方法研究等。论文的主要工作和创新如下：1．系统级标定方法研究。推导了在圆锥运动条件下IMU标定误差的误差传播公式，指出陀螺非正交安装误差与刻度因子误差是标定误差中引发圆锥误差的主要因素；引入了系统级标定方法基准坐标系的约束条件，设计了基于27维Kalman滤波器且仅以速度误差为观测量的系统级标定算法；规划了系统级标定方法的标定路径，并基于分段线性定常系统可观测性分析理论（PWCS）与奇异值分解法（SVD）对其可观测性进行了分析，提出了基于可观测度分析的加权反馈算法；仿真和实验结果表明系统级标定法的标定精度优于基于高精度姿态基准的分立标定法，陀螺非正交安装误差角标定精度优于2。2．重力扰动对高精度姿态测量系统的影响及补偿研究。进行了重力扰动影响的单通道原理性分析，指出垂线偏差是主要误差因素；推导了由垂线偏差产生的解算误差的理论表达式，并应用全球重力扰动数据库进行了仿真研究；研究了重力扰动对系统初始对准的影响，得到了垂线偏差将导致等量的初始对准姿态误差的结论；利用垂线偏差数据库对航天测量船搭载试验中在某海域的试验数据进行了补偿分析，补偿后系统精度有较大幅度提高。3.针对纯惯性姿态测量系统误差随时间积累的问题，开展了基于UKF的重力场辅助惯性姿态测量技术研究。区别于以往基于图形相关匹配算法的间接融合算法，设计了一种直接利用重力异常值作为观测量的信息融合方法，由于重力异常信息具有强非线性特征，选用UKF作为信息融合算法；提出了一种基于重力异常梯度加权的多周期反馈UKF滤波算法，仿真结果表明应用此算法后系统的位置、速度以及姿态误差均得到了较好的抑制，且算法对测量噪声与初始定位误差不敏感，与EKF算法相比无需计算局部重力场模型并对其线性化，在不同重力异常梯度下均具有较强的鲁棒性；基于航天测量船搭载试验数据开展了半实物仿真研究，结果进一步验证了本算法的可行性与适应性。4．动基座对准研究。应用PWCS理论研究了单轴旋转调制姿态测量系统动基座对准的可观测性问题，指出旋转调制能提高系统动基座对准的可观测性，降低了高精度动基座对准对载体机动的要求，研究结果表明在简单机动的情况下，单轴旋转调制姿态测量系统动基座对准的精度和速度都有所提高；针对动基座大方位失准角对准问题，推导了其非线性误差传播方程，并在此基础上设计了基于UKF的动基座非线性对准算法，仿真和实验结果均表明，UKF能很好地解决姿态测量系统动基座大方位失准角的问题。5．FIR滤波延迟补偿方法研究。针对机抖激光陀螺系统中FIR滤波器延迟将导致姿态输出实时性降低的问题，采用基于运动跟踪预测模型的Kalman滤波器，直接对有时间延迟的姿态角输出信息进行时间序列上的一步预测，得到当前时刻真实姿态角的估计值；通过对实测数据的分析，表明本方法能有效预测出运动载体的即时姿态角，提高了姿态测量系统的实时性。6．系统软硬件实现及航天测量船搭载试验。论述了基于机抖激光陀螺的高精度姿态测量系统的软硬件实现方案，研究了基于Vxworks实时操作系统的姿态测量软件开发问题，详细规划了各任务模块设计；开展了为期近7个月、行程达3万多海里的远洋航天测量船搭载试验，对论文研究的重力补偿、重力场辅助惯性测量、动基座对准可观测性、系统级标定方法进行了充分的验证分析。更多还原

【Abstract】 With the development of ring laser gyroscope’s (RLG) producing technology andthe improvement of its precision in our country, research on the attitude measurementsystem using RLG as the core component has became attractive. Taking the marinestrapdown attitude measurement system as a research object, the key factors restrictingthe improvement of attitude measurement system’s precision and the relevant solutionmethods are studied in this thesis. The content of this thesis includes systematiccalibration algorithm of inertial measurement unit(IMU), analysis of the system errorscaused by gravity disturbances and the compensation for them, gravity-aided inertialattitude measurement system, alignment on moving base, the filtering delaycompensation, the realization of system software and hardware and so on. The maincontents and contributions are as follows:1. Systematic calibration algorithm. The equations of coning errors caused bycalibration errors of IMU in the condition of a typical coning motion in strapdownsystem are derived. We draw the conclusion that the main error factors of attitudeupdating coning errors among calibration errors are gyro non-orthogonal misalignmentangles and scale factor errors; Firstly a restriction on the reference frame of RLGs andaccelerometers is added in order to make the calibration result exclusive; The systematiccalibration algorithm based on a kalman filter which includs27dimensional stateparameters is proposed, and this filter only uses the velocity errors as the observationalvectors; A programming of calibration route is proposed and its observability isanalyzed, then a weighted feedback arithmetic based on observability analysis is putforward. The simulation and experimentation indicate that the results of the systematiccalibration method are more precise than the conventional method.The verification testshows that the estimation accuracy of the gyros’ misalignment angles exceeds2.2. Research on the effects of gravity disturbances on high-precision attitudemeasurement system and its compensation method. The single channel theoreticalanalysis is carried out and it is pointed out that the vertical deflection is the primaryfactor; The theoretical expression of solution errors caused by vertical deflection isderived, and then a simulation based on global gravity disturbances database is made,which proves the accuracy of the theoretical analysis;The effects of gravity disturbanceson initial alignment are researched and we draw a conclusion that the vertical deflectionwill lead to equal magnitude initial attitude errors;The experimentation carried out in anoceangoing survey ship indicates that the precision can be improved obviously thanbefore, after compensated using global vertical deflection database.3. In order to restrain the increase of errors of the pure inertial attitude measurementsystem, a method that uses gravity information assisted inertial measurement system isproposed. A direct information fusion algorithm for gravity-aided inertial measurement system is designed, whereas the gravity anomaly information has strong nonlinearility,so UKF is selected as the information fusion algorithm. A weighted and multicyclefeedback UKF arithmetic based on gravity anomaly grads is put forward, and thesimulation results indicate that the proposed algorithm can improve the precision inposition, velocity and attitude greatly, and this algorithm is not sensitive to initialposition errors and measurement noise. It does not need the analytical local gravity fieldmodeling and linearization compared with EKF(expanded kalman filter), and itsrobustness is strong in different gravity anomaly gradients. Furthermore, ahardware-in-the-loop simulation based on the data of oceangoing survey ship isaccomplished, and the results validate the algorithm’s feasibility and adaptability.4. Research on the initial alignment on moving base. The observability duringinitial alignment of strapdown single-axis rotating attitude measurement system onmoving base is studied by the PWCS method. The results demonstrate that theadditional motion is not necessary for rotating system to improve alignment precision.In order to improve the performance of initial alignment under large azimuthmisalignment and on moving base, a detailed derivation of nonlinear error models underthese conditions is proposed and a nonlinear alignment arithmetic based on UKF isdesigned. The results show that the UKF arithmetic is a appropriate method for initialalignment under large azimuth misalignment.5. Study on the FIR filtering delay compensation method. A filtering delaycompensator which is based on attitude movement tracking model and kalman filter isdesigned. The compensator can obtain the estimation of the real attitude angles bypredicting the current output attitude results for a certain time.The experiments showthat, this method can estimate and predict the carrier attitude effectively and cancompensate the adverse effect caused by FIR filter delay.6. Research on the realization of the hardware&software and experiments boardedon oceangoing survey ship.The realization of software&hardware is discussed, while adetailed design planning of all task modules of attitude measurement software based onVxworks real-time operating system is put forward; The5months’ system experimentsboarded on the ship were completed and the voyage reached up to20000nautical miles,the verification and analysis on the gravity compensation, gravity-aided inertialmeasurement, initial alignment on moving base, and systematic calibration algorithmwere carried out adequately during the voyage.更多还原