导航系统中的多径误差抑制算法研究

发布时间：2018-05-19 22:47

本文选题：参数估计 + 粒子滤波　；参考：《太原理工大学》2017年硕士论文

【摘要】：全球定位系统(Global Position System,GPS)在人们的生活中被广泛应用,而随着人们对导航系统中高精度定位需求的日益增长,干扰抑制成为了研究热点。在众多影响定位精度的因素中,多径干扰是降低定位精度的主要误差源之一。抑制多径干扰造成的误差,其难点在于多径干扰具有位置上的不相关性、时间上的不确定性,不能通过现有的差分技术来消除。而基于数据处理的多径误差抑制方法符合目前软件接收机的发展趋势,因此,本文重点研究了基于参数估计的多径误差抑制算法,旨在通过数据处理的方法估计多径参数,并依据所估参数重构多径信号,进而消除多径干扰的影响,以此来达到抑制多径误差的目的。本文重点研究了高斯噪声和非高斯噪声下的多径误差抑制算法。扩展卡尔曼滤波(Extended Kalman Filter,EKF)和粒子滤波(Particle Filter,PF)分别是高斯噪声和非高斯噪声下用于多径估计的两种典型算法,取得了很多研究者的关注。但在高斯噪声下,基于EKF的多径估计算法仍存在如下问题:对初值敏感、在对非线性方程进行线性化过程中会产生截断误差,致使估计结果在真值附近具有较大的波动。在非高斯噪声下,PF算法虽然应用广泛,滤波效果较好,但标准的粒子滤波在进行参数估计时存在粒子枯竭的问题,致使新产生粒子的多样性减少,降低了PF的参数估计精度。针对EKF存在对初值敏感、滤波结果波动较大的问题,本文提出一种基于PF和滑动平均EKF的多径估计算法。该算法在运行的初始阶段,首先利用PF估计多径参数,然后将得到的多径参数的粗略估计值作为EKF的初始估计值,以解决EKF对初值敏感的问题。接着利用EKF进行算法的后续估计,并对EKF的估计结果进行滑动平均,最后将滑动平均后的滤波结果作为多径参数的估计结果。仿真结果表明,改进后的多径估计算法相比EKF和PF具有更优的估计性能,可有效降低估计结果的波动幅度,同时克服了EKF对初值敏感的问题。针对标准粒子滤波存在的粒子枯竭问题,本文提出一种基于自适应差分进化的粒子滤波(Adaptive Differential Evolution Particle Filter,ADE-PF)算法,该算法利用自适应差分进化算法代替PF中的重采样策略来产生新粒子,使粒子朝着状态后验概率密度函数的高似然区移动,同时提高了粒子的多样性。所采用的ADE算法,通过一种非线性自适应调节策略来自适应地调整缩放因子和交叉因子,以提高改进PF中DE(Differential Evolution)优化部分的寻优能力。为了验证该算法的有效性,分别在高斯噪声环境和非高斯噪声环境下,将所提出的ADE-PF算法应用于多径估计。通过仿真验证了ADE-PF算法可克服标准PF存在的粒子枯竭问题;与PF、EKF和DE-PF相比,ADE-PF算法具有更优的多径估计性能。本文研究内容为山西省自然科学基金(No.2014021022-7)的重要组成部分,为高斯噪声和非高斯噪声下的多径误差抑制算法研究提供了参考,对提高导航系统的定位精度有重要的理论意义和广泛的应用前景。
[Abstract]:Global Position System (GPS) is widely used in the people's life. With the increasing demand for high precision positioning in the navigation system, interference suppression has become a hot topic. Among the many factors affecting the positioning accuracy, multipath interference is one of the main error sources to reduce the positioning accuracy. The difficulty of the error caused by interference is that the multipath interference is not related to the position, the time is uncertain and can not be eliminated by the existing differential technology. And the method of multipath error suppression based on data processing is in line with the development trend of the current software receiver. Therefore, this paper focuses on the multipath error based on parameter estimation. The difference suppression algorithm is designed to estimate multipath parameters by data processing, and to reconstruct multipath signals based on the estimated parameters and eliminate the influence of multipath interference in order to suppress multipath error. This paper focuses on the multi-path error suppression algorithm under Gauss noise and non Gauss noise. Extended Calman filter (Extended K). Alman Filter, EKF) and particle filter (Particle Filter, PF) are two typical algorithms for multipath estimation under Gauss noise and non Gauss noise, which have obtained many researchers' attention. But under Gauss noise, the following problems still exist in the multipath estimation algorithm based on EKF, which is sensitive to the initial value and linearized the nonlinear equation. There will be a truncation error in the process, which leads to the larger fluctuation of the estimated results near the true value. Under the non Gauss noise, the PF algorithm is widely used and the filtering effect is good, but the standard particle filter has the problem of particle exhaustion in the parameter estimation, which reduces the diversity of the newly generated particles and reduces the estimation accuracy of the parameters of the PF. In this paper, a multipath estimation algorithm based on PF and sliding mean EKF is proposed to solve the problem that the EKF is sensitive to the initial value and the filtering results are very volatile. In the initial stage of the operation, the algorithm first uses PF to estimate the multipath parameters and then the rough estimation value of the obtained multipath parameters as the initial value of the EKF, so as to solve the initial value sensitivity of the EKF. Then the EKF is used to carry out the subsequent estimation of the algorithm, and the estimation results of the EKF are sliding averaging. Finally, the filtering results after the sliding average are used as the estimation results of the multipath parameters. The simulation results show that the improved multipath estimation algorithm has a better estimation performance than the EKF and the PF, and can effectively reduce the fluctuation of the estimated results. At the same time, the problem of EKF sensitivity to the initial value is overcome. Aiming at the problem of particle exhaustion in the standard particle filter, an adaptive differential evolution based particle filter (Adaptive Differential Evolution Particle Filter, ADE-PF) algorithm is proposed. The algorithm uses the self adaptive differential evolution algorithm to replace the resampling strategy in PF. New particles are generated to move the particle to the high likelihood region of the posterior probability density function and improve the diversity of the particles. The ADE algorithm used to adjust the scaling factor and the cross factor adaptively by a nonlinear adaptive adjustment strategy to improve the optimization of the optimized part of the improved PF (DE) (Differential Evolution). Ability. In order to verify the effectiveness of the algorithm, the proposed ADE-PF algorithm is applied to multipath estimation in Gauss noise environment and non Gauss noise environment. It is verified by simulation that the ADE-PF algorithm can overcome the problem of particle exhaustion in standard PF; compared with PF, EKF and DE-PF, the ADE-PF algorithm has a better performance of multipath estimation. The research content is an important part of the natural science foundation of Shanxi (No.2014021022-7). It provides a reference for the study of the multipath error suppression algorithm under Gauss noise and non Gauss noise. It has important theoretical significance and wide application foreground for improving the positioning accuracy of the navigation system.
【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P228.4

【参考文献】