惯性行人导航系统的算法研究

发布时间：2018-11-24 14:09

【摘要】：常见的导航系统,在导航卫星或地面信标失效的环境中存在导航盲目。惯性导航系统作为一种自主式的导航系统,具有良好的隐蔽性、稳定性和抗干扰性,可以全天时、全天候、全地域地工作,近年来成为行人导航领域的一大研究热点。然而,纯惯性导航具有导航误差不断积累的特点,需设计行之有效的误差校正方法来提高系统的长期导航精度。本文基于小尺寸、低成本的微惯性传感器,结合足部运动的周期特性,开展了将零速修正辅助的惯性导航技术应用于行人导航的相关算法研究。主要研究内容如下：1、研究了静基座情况下捷联惯导系统的初始对准技术。初始对准是惯性导航系统的关键技术之一,其主要目的是确定系统的初始条件。本文基于欧拉平台误差角的概念描述了理论导航坐标系到计算导航坐标系之间的失准角,推导了捷联惯导系统在不同失准角情况下的系统误差模型,为文中误差校正算法的研究奠定了理论基础。2、研究了基于动态阈值和聚类分析的步态检测方法。现有的步态检测方法由于没有充分合理地考虑测量值波动对步态检测的影响,大都存在检测结果不准确和对检测参数敏感等不足,进而影响后续零速修正方法的正确性和有效性。本文通过详细分析各个检测参数的作用和它们之间的耦合关系,提出了一种基于聚类分析的自适应步态检测方法,该方法是一种改进的平区检测方法,能够克服现有检测方法的诸多不足,并扩大检测方法可行的参数空间,进而提高步态检测的精确性和可靠性。3、研究了基于卡尔曼滤波器的导航误差估计方法。在进行零速修正时,卡尔曼滤波算法可以充分利用速度误差与姿态误差、位置误差之间的耦合关系,估计和校正更多的导航误差。本文为了降低滤波过程的计算量和高阶截断误差、避免引入额外的建模误差并减小滤波发散的可能性,利用间接滤波原理对原始系统方程进行了分解,然后结合行人导航的具体应用,对分解后的误差子系统进行了简化,通过分析支撑相内零速度观测值的可靠性,滤波过程中未将传感器误差作为增广的状态向量进行建模和估计。4、研究了基于固定区间平滑器的导航误差估计方法。卡尔曼滤波算法仅能对支撑相内的导航误差进行估计,容易在摆动相过渡到支撑相时引起导航信息的突变。固定区间平滑算法能够对整个步态周期内的导航误差进行估计,实现步态时相的平稳过渡,从而提高系统的精确性和稳定性。为了使平滑算法满足在线运行的需求,本文在不降低系统性能的情况下,设计了未对位置误差和传感器误差进行建模和估计的降阶平滑滤波算法,以步进的方式进行导航误差的估计、平滑和校正,从而达到准实时的平滑估计效果。
[Abstract]:The common navigation system, in navigation satellite or ground beacon failure environment, navigation blind. Inertial navigation system, as an autonomous navigation system, has good concealment, stability and anti-interference. It can work all day, all weather, and all over the area. In recent years, inertial navigation system has become a research hotspot in the field of pedestrian navigation. However, pure inertial navigation has the characteristics of continuous accumulation of navigation errors, it is necessary to design an effective error correction method to improve the long-term navigation accuracy of the system. Based on the micro inertial sensor with small size and low cost, combined with the periodic characteristics of foot motion, this paper studies the application of zero-velocity correction assisted inertial navigation technology to pedestrian navigation. The main contents are as follows: 1. The initial alignment technology of strapdown inertial navigation system under static base is studied. Initial alignment is one of the key technologies of inertial navigation system, and its main purpose is to determine the initial conditions of the system. Based on the concept of error angle of Euler platform, this paper describes the misalignment angle between the theoretical navigation coordinate system and the computational navigation coordinate system, and deduces the systematic error model of the strapdown inertial navigation system under different misalignment angles. It lays a theoretical foundation for the research of error correction algorithm in this paper. 2. The gait detection method based on dynamic threshold and clustering analysis is studied. Because the existing gait detection methods do not fully and reasonably consider the influence of measurement value fluctuation on gait detection, most of the existing gait detection methods have some shortcomings, such as inaccurate detection results and sensitivity to detection parameters, etc. Furthermore, the correctness and effectiveness of the subsequent zero-speed correction method are affected. In this paper, an adaptive gait detection method based on clustering analysis is proposed by analyzing the function of each detection parameter and the coupling relationship between them in detail. This method is an improved flat area detection method. It can overcome the shortcomings of the existing detection methods and expand the feasible parameter space of the detection methods, and then improve the accuracy and reliability of gait detection. 3. A navigation error estimation method based on Kalman filter is studied. In the zero-velocity correction, the Kalman filtering algorithm can make full use of the coupling relationship between velocity error, attitude error and position error, and estimate and correct more navigation errors. In order to reduce the computational complexity and high order truncation error in the filtering process, to avoid the introduction of additional modeling errors and to reduce the possibility of filtering divergence, the original system equations are decomposed by indirect filtering principle. Then combining with the concrete application of pedestrian navigation, the decomposed error subsystem is simplified, and the reliability of zero velocity observation in support phase is analyzed. The sensor error is not modeled and estimated as an augmented state vector in the filtering process. 4 the navigation error estimation method based on fixed interval smoother is studied. The Kalman filtering algorithm can only estimate the navigation error in the support phase, and it is easy to cause the sudden change of navigation information when the swing phase is transitioned to the supporting phase. The fixed-interval smoothing algorithm can estimate the navigation error in the whole gait cycle and realize the steady transition of the gait phase, thus improving the accuracy and stability of the system. In order to make the smoothing algorithm meet the requirements of on-line operation, a reduced order smoothing filtering algorithm without modeling and estimating the position error and sensor error is designed without reducing the performance of the system. The error estimation, smoothing and correction of navigation error are carried out step by step to achieve the effect of quasi-real-time smoothing estimation.
【学位授予单位】：大连理工大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TN713

【相似文献】