【作者】 姬忠校；

【导师】 马彩文；

【作者基本信息】 中国科学院研究生院（西安光学精密机械研究所） ， 光学工程， 2011， 博士

【摘要】 陀螺仪作为惯性制导、导航以及姿态控制的核心器件已经发展了三代。第一代机电式陀螺由于其性价比和寿命等方面上的缺陷，正逐渐被以光学Sagnac效应为基础的光学陀螺所取代。第二代环形激光陀螺在精度和可靠性方面取得了巨大的进步，使得其在惯性导航系统中得到了广泛的应用，然而为解决“闭锁”问题而加入的机械抖动装置使得其系统变得非常复杂，由此产生了第三代陀螺技术——光纤陀螺。光纤陀螺以其小体积、高性价比、精度适用范围广、无运动部件等优点，目前在中低精度的陀螺应用领域中得到了广泛地使用。干涉式光纤陀螺作为光纤陀螺技术的一种，其基础理论和技术虽已成熟，然而我国光纤陀螺技术的发展水平和精度等级与国外发达国家相比仍有巨大的差距；同时由于受到国内陀螺光学器件生产工艺和技术水平的限制，目前改善陀螺性能、提高陀螺的精度等级主要依靠电子学和信号处理的技术手段，因此研究如何在现有的技术条件下改善光纤陀螺性能指标的相关技术，对于光纤陀螺的技术发展和工程实践具有重要的理论意义和经济价值。本研究课题以干涉式数字闭环光纤陀螺作为研究的对象，对改善陀螺性能的相关技术开展研究工作。重点研究了评价陀螺性能的零偏稳定性、角随机游走系数以及标度因数三个关键技术指标的影响因素，以及改善这三个指标的技术措施。总结本课题论文的工作内容和创新成果可以概括如下：（1）根据干涉式光纤陀螺的工作原理以及输出信号的特性，建立了数字式闭环光纤陀螺的动态模型和随机模型；采用数字信号仿真的方法对光纤陀螺存在的五种主要噪声类型进行仿真实验，通过Allan方差分析对比真实的陀螺输出数据，验证了模型和随机噪声仿真的有效性；（2）研究了造成闭环光纤陀螺标度因数误差的主要因素，以及标度因数误差补偿的技术方案，并通过模型仿真的方法验证了基于四态方波调制增益误差补偿方案的有效性；（3）研究了影响光纤陀螺角随机游走系数以及信噪比的主要因素；针对陀螺光源相对强度噪声限制信噪比提高的问题，研究了目前广泛采用三种抑制相对强度噪声的技术方案以及缺陷，并提出了一种基于RLS自适应噪声对消技术。仿真结果表明该技术对噪声传输延迟和幅值变化具有很强地适应能力，对陀螺的零偏稳定性和角随机游走系数指标有较大改善；（4）根据光纤陀螺非线性、非平稳性的信号特点，研究对比了LMS自适应滤波算法在不同调整步长和滤波器阶数下的滤波效果，从对比的实验结果中得出了针对光纤陀螺信号特点的LMS滤波器参数的确定原则；针对LMS自适应滤波算法的缺陷，提出了基于非参数、可变步长的归一化LMS(NVSS-NLMS)滤波算法的技术方案，实验结果表明该技术方案相比基于LMS算法的技术方案具有更好的滤波效果；（5）针对自适应滤波算法需要确定适合的参考信号问题，研究了基于信号本身蕴含的、反映不同时间尺度下信号局部特征的经验模分解（EMD）滤波技术，并针对陀螺动态信号EMD分解后如何重建信号的问题，提出了基于序列均方误差检测的EMD滤波方法，该方法为信号重建提供了客观的判断标准。通过对陀螺信号的滤波处理，陀螺的零偏稳定性和角随机游走系数指标得到了显著改善。更多还原

【Abstract】 A gyroscope as the core components of the inertial guidance, navigation andattitude has been developed three generations. The first generation of electromechanicalgyro due to it’s these disadvantages of performance to cost ratio, life-time and so on, isgradually being replaced by the optical gyros based on optical Sagnac effect. Thesecond generation of ring laser gyro (RLG) has made great process in the aspect ofprecision and reliability, which makes it to be widely used in the inertial navigationsystem. However, to solve “locking” phenomenon and added the mechanical jitterdevice in RLG, it makes RLG system to become very complex. This results in the thirdgeneration of gyro technology-Fiber Optic Gyroscope (FOG). FOG has theseadvantages of small size, low cost, wide range of precision, no moving part and so forth,which is widely used in the low and middle precision gyro applications currently.The interferometric FOG is a kind of FOG technology, and its fundamental theoryand technology have fully matured. But there is a huge gap compared the domestictechnology with developed countries in the developing level and precision of FOG. Dueto limitations of the production technologic level of the optical devices, the presenttechnologic measures rely mainly on the electronics and signal processing techniquesfor improving performance and enhancing the precision of FOG. So with the actualtechnical conditions the research of the relative technology about how to improve theperformance of FOG, has important theoretical significance and economic value for thetechnical development of FOG and engineering practice.In this dissertation the interferometric digital closed-loop FOG is as the object ofresearch. The research work about the improved performance-related technology ofFOG is to be carried out. The effecting factors and the improved performancetechnology of the three key indexes, Bias Stability, Angle Random Walk Coefficient andScale Factor, are researched mainly, which is employed to evaluate the performance ofFOG. In this article the main content and innovations can be summarized as follows:(1) According to the operation principle and the output signal characteristics of FOG, thedigital closed-loop fiber optic gyro’s dynamic model and random model areestablished. The main five types of random noise in FOG are simulated with themethod of digital signal simulation. The random model and the simulation methodof noise are verified by Allan variance analysis and comparison of the real gyro’soutput data. (2) To study the main factors which causes Scale Factor error and the compensationtechnologies of Scale Factor error are introduced. The effectiveness of the gain errorcompensation technology based on the four-state square-wave modulation is provedwith simulation of the model.(3) To study the main factors of impacting Angle Random Walk Coefficient and SNR ofFOG. To solve the problem of the source Relative Intensity Noise (RIN) limiting toenhance SNR of FOG, the three suppression RIN technologies and their drawbacksare studied and a kind of the adaptive noise cancellation technology based on RLSalgorithm is proposed in this paper. The simulation shows that this technology has astrong ability to meet the change of the transfer delay and amplitude of RIN and theFOG’s Bias Stability and Angle Random Walk Coefficient have been improved in acertain extent after using this technology.(4) According to the FOG’s signal characteristics of nonlinearity and nonstationarity, tostudy and compare the LMS adaptive filter algorithm at different step size and theorders of filter. The principle to determining the LMS filter’s parameters result fromthe experimental results. To avoid the drawbacks of LMS filter algorithm, thefiltering technology based on Non-parametric, Variable Step Size Normalized LMS(NVSS-NLMS) algorithm is proposed. The experiment shows that this technologycompared to LMS algorithm has a better filtering effect.(5) To avoid how to choice the appropriate reference signal for the adaptive filteringalgorithm, the Empirical mode decomposition (EMD) filtering technology isresearched, which based on different time scale of the local characteristics in signalitself inclusively. To solve how to reconstructed signal for EMD, a method based onsequence Mean Square Error (MSE) detection is proposed, which provides anobjective criterion for signal reconstruction of EMD filtering method. By filteringthe signal of FOG, the index of Bias Stability and Angle Random Walk Coefficientare enhanced in a great extent.更多还原