高精度MEMS陀螺仪的滤波算法研究

发布时间：2018-10-29 21:17

【摘要】：随着惯性导航技术的快速发展,MEMS陀螺仪的应用也日益广泛起来。如何提高MEMS陀螺仪输出数据的精度,已成为国内外专家学者研究的热点问题。针对这一问题,本文在基于信号采集系统的基础上,采集了陀螺仪输出的原始数据,并对其进行了误差分析,然后基于时间序列分析方法建立了陀螺仪随机漂移的数学模型。最后,应用不同的滤波技术对陀螺仪的零点漂移与动态误差进行了研究与仿真分析,从而确定了最终的信号处理方法。首先,本文介绍了信号采集系统的基本原理,并进一步分析了其硬件结构与软件组成,然后设计了一种以STM32为核心控制器的信号采集系统。接着对MEMS陀螺仪的工作原理作了详细说明,同时分析了其零点调整机制。另外,确定了陀螺仪原始信号的采样频率,并采集了陀螺仪的原始输出数据,接着在MATLAB环境下采用Allan方差分析法对其误差项进行了辨识,并确定了其各误差项的系数。其次,本文在深入理解平稳随机过程统计特性的基础上,对陀螺仪的输出数据进行了必要的预处理,并确定用时间序列分析法建立陀螺仪随机漂移的误差模型,同时确定了模型定阶以及参数估计的方法。然后,本文分析了传统卡尔曼滤波器与二阶低通滤波器的设计思想、工作原理以及数学模型,接着对陀螺仪输出数据的统计特性进行了检验,并经过实验确定了其误差模型的阶数与参数,最终以AR(1)模型作为其误差模型。最后分别使用卡尔曼滤波器与二阶低通滤波器在MATLAB环境下对陀螺仪的静态数据进行了仿真分析,通过分析比较滤波结果,确定用传统的卡尔曼滤波器稳定零点。最后,本文分析了带有确定性输入项的卡尔曼滤波器的工作原理与数学模型,接着用二阶低通滤波器处理陀螺仪的动态数据,分析实验结果发现,虽然可以在一定程度上减轻误差,但是存在着滤波滞后的问题,因此本文设计了一种新的算法,即扩展的卡尔曼滤波器与二阶低通滤波器融合的算法,通过对二阶滤波的结果进行差分,然后把差分结果作为卡尔曼滤波器的确定性输入,实验结果表明:该算法可以很好的处理陀螺仪的动态误差。
[Abstract]:With the rapid development of inertial navigation technology, the application of MEMS gyroscope is becoming more and more extensive. How to improve the precision of output data of MEMS gyroscopes has become a hot issue for experts and scholars at home and abroad. In order to solve this problem, based on the signal acquisition system, the original data of gyroscope output are collected, and the error is analyzed. Then, the mathematical model of random drift of gyroscope is established based on the time series analysis method. Finally, the zero drift and dynamic error of gyroscope are studied and simulated by using different filtering techniques, and the final signal processing method is determined. Firstly, this paper introduces the basic principle of the signal acquisition system, and further analyzes its hardware structure and software composition, then designs a signal acquisition system with STM32 as the core controller. Then the working principle of MEMS gyroscope is explained in detail, and its zero adjustment mechanism is analyzed. In addition, the sampling frequency of the original signal of the gyroscope is determined, and the original output data of the gyroscope are collected. Then, the error terms are identified by the Allan variance analysis method under the MATLAB environment, and the coefficients of each error term are determined. Secondly, on the basis of deeply understanding the statistical characteristics of stationary stochastic process, this paper preprocesses the output data of gyroscope and establishes the error model of random drift of gyroscope by time series analysis. At the same time, the order of the model and the method of parameter estimation are determined. Then, the design idea, working principle and mathematical model of the traditional Kalman filter and the second-order low-pass filter are analyzed, and the statistical characteristics of the output data of the gyroscope are tested. The order and parameters of the error model are determined by experiments. Finally, the AR (1) model is used as its error model. Finally, the static data of gyroscope are simulated by using Kalman filter and second-order low-pass filter in MATLAB environment, and the stable zero point of gyro is determined by comparing the filtering results with the traditional Kalman filter. Finally, the working principle and mathematical model of Kalman filter with deterministic input term are analyzed, and the dynamic data of gyroscope are processed by second-order low-pass filter. Although the error can be reduced to a certain extent, there is the problem of filtering lag. Therefore, a new algorithm is designed in this paper, which is the fusion algorithm of extended Kalman filter and second-order low-pass filter. Through the difference of the second order filter result and the difference result as the deterministic input of the Kalman filter, the experimental results show that the algorithm can deal with the dynamic error of the gyroscope very well.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN96;TN713

【参考文献】