低成本MEMS陀螺仪随机漂移误差的建模及修正

发布时间：2018-05-11 02:04

本文选题：MEMS陀螺仪 + 随机漂移误差　；参考：《西南大学》2017年硕士论文

【摘要】：近年来,MEMS陀螺仪作为惯性导航技术中十分重要的部分,由于其具有成本低、尺寸小、重量轻、集成度高等一系列优点,在惯性导航、工业控制及电子产品等领域得到广泛的应用。尽管与传统类型陀螺仪相比,MEMS陀螺仪具有诸多优势,但由于制造工艺和设计水平等原因,其测量精度相对较低,往往无法满足实际使用需求,极大的制约了MEMS陀螺仪的发展和应用。尤其对于陀螺仪的随机漂移误差,由于其形成的机理非常复杂,没有明确的规律且随外界环境变化而变化,不能用常规的方法补偿修正,并且无法完全消除,是限制MEMS陀螺精度提高的主要因素。有鉴于此,本文以目前较常用的低成本MEMS惯性器件MPU-6050中的陀螺仪为研究对象,开展了对MEMS陀螺仪随机漂移误差建模分析与修正技术的研究,本文的主要研究内容及取得结论如下:首先,为采集陀螺仪的静态漂移数据,以STM32F103C8T6为核心处理器设计并构建了MEMS陀螺仪误差数据采集系统,实现了对MPU-6050中陀螺仪静态漂移数据的采集。其次,在对MEMS陀螺仪随机漂移误差建模与滤波技术的研究现状与发展动态调研分析的基础上,研究了MEMS陀螺仪静态漂移误差产生的机理及随机漂移误差的组成,并采用Allan方差法辨识已采集到的MEMS陀螺仪静态漂移误差。误差分析结果表明,该陀螺仪的静态漂移误差主要由零偏不稳定性、速率随机游走及速率斜坡三部分噪声构成。再次,对MEMS陀螺仪静态漂移数据进行预处理和数据检验以得到平稳随机的误差数据,然后采用AIC准则对模型定阶并用Yule-Walker方程确定模型参数,在此基础上建立陀螺仪随机漂移误差的时间序列模型,结合误差模型设计了卡尔曼滤波器,并对误差数据滤波。对比滤波前后各项指标,经卡尔曼滤波后的MEMS陀螺仪随机漂移误差数据的方差下降为滤波前的11.7%,影响MEMS陀螺仪精度的主要随机误差中,零偏不稳定性噪声系数减少了68.6%,速率随机游走噪声系数减少了67.7%,速率斜坡噪声系数减少了68.0%,表明卡尔曼滤波能有效的减少MEMS陀螺仪随机漂移误差。最后,针对时间序列和卡尔曼滤波对MEMS随机漂移数据处理中的不足,采用Singer运动模型建立MEMS随机漂移误差的模型,结合粒子-卡尔曼组合滤波的方法处理零均值化后的陀螺随机漂移误差数据。比较滤波前后各项指标,经粒子-卡尔曼组合滤波后误差数据的方差下降为滤波前的1.2%,影响MEMS陀螺仪精度的主要随机误差中,零偏不稳定性噪声系数减少了72.8%,速率随机游走噪声下降了76.2%,速率斜坡噪声下降了74.6%,表明粒子-卡尔曼组合滤波的方法能显著的抑制MEMS陀螺随机漂移误差。研究结果表明,本文采用的时间序列建模与卡尔曼滤波、基于Singer模型建模与粒子-卡尔曼滤波组合滤波两种方案均能有效的抑制MEMS陀螺仪的随机漂移误差,且粒子-卡尔曼组合滤波方法的滤波效果要优于卡尔曼滤波。研究成果能有效修正MEMS陀螺仪的随机漂移误差,对于提高以低成本MEMS陀螺仪作为主要惯性传感器的惯性导航系统的导航精度具有一定的实用价值。
[Abstract]:In recent years, as a very important part of inertial navigation technology, MEMS gyroscope has been widely used in the fields of inertial navigation, industrial control and electronic products because of its advantages such as low cost, small size, light weight and high integration. Compared with the traditional type gyroscope, MEMS gyroscope has many advantages, but it has a lot of advantages. The measurement precision of the manufacturing process and the design level is relatively low, which often can not meet the actual use demand and greatly restricts the development and application of the MEMS gyroscope. Especially for the gyro random drift error, because the mechanism is very complex, there is no definite law and changes with the external environment, and it can not be used. It is the main factor to limit the precision of MEMS gyroscope, which is the main factor to limit the precision of the gyroscope. In this paper, the research on the modeling analysis and correction of the random drift error of the MEMS gyroscope is carried out in this paper, which is the research object of the gyroscope in the low cost MEMS inertial device MPU-6050. The research content and the conclusions are as follows: first, in order to collect the static drift data of the gyroscope, the MEMS gyroscope error data acquisition system is designed and constructed with STM32F103C8T6 as the core processor. The data collection of the gyro static drift in the MPU-6050 is realized. Secondly, the modeling and filtering of random drift error of the MEMS gyroscope is made. On the basis of research status and development dynamic investigation and analysis, the mechanism of the static drift error of MEMS gyroscope and the composition of random drift error are studied. The static drift error of the MEMS gyroscope has been identified by the Allan variance method. The error analysis results show that the static drift error of the gyroscope is mainly caused by the zero bias instability. Three parts of the rate random walk and the rate slope are made up. Again, the static drift data of the MEMS gyroscope are preprocessed and the data are tested to get the stationary random error data. Then the model parameters are determined by the AIC criterion and the model parameters are determined by the Yule-Walker equation. On this basis, the time of random drift error of the gyroscope is established. The Calman filter is designed with the error model, and the error data filter is designed. The variance of the random drift error data of the MEMS gyroscope after the Calman filter is reduced to 11.7% before and after the Calman filtering. In the main random error that affects the precision of the MEMS gyroscope, the zero bias instability noise coefficient is reduced. 68.6%, the rate of random walk noise reduction is reduced by 67.7% and the rate of rate slope noise reduction is reduced by 68%. It shows that Calman filter can effectively reduce the random drift error of MEMS gyroscope. Finally, the random drift of MEMS drift data is not enough for time series and Calman filtering, and the random drift of MEMS is established by using the Singer motion model. The error model is combined with the particle Calman combination filtering method to deal with the random drift error data of the gyroscope after the zero mean. The variance of the error data is reduced to 1.2% before filtering after the particle Calman combination filtering, and the zero bias instability noise is affected by the main random error of the precision of the MEMS gyroscope. The sound coefficient is reduced by 72.8%, the rate of random walk noise is reduced by 76.2% and the rate of rate slope is reduced by 74.6%. It shows that the method of particle Calman combination filtering can significantly inhibit the random drift error of MEMS gyro. The results show that the time series modeling and Calman filtering adopted in this paper are based on the Singer model modeling and particle Carle. Two schemes of Mann filter combined filter can effectively suppress random drift error of MEMS gyroscope, and the filtering effect of particle Calman combination filtering method is better than Calman filter. The research results can effectively modify the random drift error of MEMS gyroscope, and improve the inertia of the low current MEMS gyroscope as the main inertial sensor. The navigation accuracy of the sex navigation system is of practical value.

【学位授予单位】：西南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN96

【参考文献】