Research On Fusion Positioning Method Based On Indoor Geomagnetic Field And MEMS Inertial Sensor

With the development of society and economy,people's positioning needs are changing from outdoor to indoor.Global satellite navigation systems cannot receive enough satellites signals indoor environments and cannot provide location services.The current indoor positioning technologies are endless,but most positioning technologies require the installation of signal transmitters in the positioning area,which undoubtedly increases the positioning cost.The geomagnetic indoor positioning technology and PDR based on the mobile phone do not need to install any infrastructure,only use indoor geomagnetic field and MEMS inertial sensor to realize position estimation.The indoor geomagnetic field is ubiquitous and remains stable over time,but the degree of fingerprint position discrimination is low.PDR technology can provide accurate position estimation in short time,but there is a phenomenon of error accumulation.This paper makes full use of the advantages of indoor geomagnetic field and PDR technology,and design a fusion positioning algorithm based on indoor geomagnetic field and MEMS inertial sensors,aiming to provide low-cost,high-precision indoor positioning services.The research content and results of this article are as follows:(1)The process of extracting multi-dimensional geomagnetic positioning features is introduced,and its positioning characteristics are analyzed in terms of stability and location differences.The source of the magnetometer error is analyzed in detail,the magnetometer error model is established,and a method based on least squares ellipsoid fitting is proposed to calibrate the magnetometer.Experiments show that the algorithm can effectively eliminate the soft iron error and hard iron error of the magnetometer;The improved Butterworth low pass filter is used to denoise the geomagnetic data to ensure that the data is not distorted while filter the noise.(2)To address the mismatching phenomenon of single-point geomagnetic matching,a geomagnetic positioning algorithm based on local geomagnetic variance is proposed.The local fingerprint database is determined by using PDR to estimate the position and step length.The variances of geomagnetic features of each dimension in the local fingerprint database were taken as the weight of single point matching results of each dimension.Aiming at the problem of a small number of jumping points in the sequence matching result,the single-point geomagnetic matching result is combined with the MD-DTW algorithm to improve the accuracy of sequence matching.Experiments show that the average positioning accuracy of improved sequence matching is about 1m.(3)Improve the accuracy of PDR.Improve peak and valley detection algorithm is used to improve the accuracy of step counting;a linear combination model of step length is proposed to improve the applicability of step length estimation;the accuracy of different heading estimation algorithms is tested based on Android platform,and the experiment shows the accuracy and real-time performance of heading estimation using rotating vector sensors are higher.(4)The fusion method of indoor geomagnetic filed and MEMS inertial sensor is proposed.Aiming at the particle degradation and depletion problems in traditional particle filter algorithms,the unscented particle filter algorithm is used to fuse the geomagnetic positioning with the PDR positioning results,and the particles are sampled in the error ellipse determined by the unscented Kalman filter result.It shows that the algorithm can effectively reduce the degradation and depletion of particles,and the average positioning error is within 1m.This thesis has 55 pictures,5 tables and 99 references.