Research On Technology Of GPS/MIMU Deep Integration

GPS/INS Deeply Integrated(DI)technology has been a hotspot in integrated navigation research currently.The research of GPS/INS DI technology is helpful to grasp the development trend of integrated navigation technology,master the core technology and break the technological monopoly by developed countries.The DI navigation system based on GPS and MEMS-IMU(MIMU)can provide high positioning accuracy,good dynamic performance,and strong anti-jamming capability.Moreover,it has many other advantages such as low-cost,low power consumption and easily integrated development,making it an important topic in the field of DI navigation.As GPS/MIMU DI technology mainly involves the MEMS inertial navigation technology,GPS receiver technology and data fusion technology,this study mostly focuses on four aspects: stochastic errors analysis of MEMS inertial sensors,MIMU-aided GPS signal tracking,vector tracking,and DI system design.The principal work is as follows:1.To overcome the drawbacks that Allan variance cannot identify time-varying characteristics of MIMU stochastic errors,this paper takes advantages of DAVAR that can reflect the time-varying characteristics of noises and combines it with Allan variance,thus achieving the purpose of analyzing the time-varying characteristics of MIMU stochastic errors.Since the traditional identifying methods of MIMU stochastic errors require the storage of a large amount of data and offline calculation,research on identification algorithms of online MIMU stochastic errors is carried out.As the equivalent ARMA method has a complex modeling process and cannot effectively identify the quantization noise,this study starts with the mathematic model of Allan variance to directly build a nonlinear state space model for MIMU stochastic errors,thereby to realize the online estimation of MIMU stochastic errors with the nonlinear filtering method.As a result,the validity of DAVAR and online estimation algorithm has been verified through simulation with the real data of ADIS16405 MEMS gyro.2.As for the error characteristics of MIMU-aided GPS tracking loop,this paper firstly introduces DLL tracking error and PLL tracking error in detail.Then taking PLL as an example,this research starts from the mathematic model and physical structure toanalyze the design ofMIMU-aided PLL.As the minimum tracking error at the optimum bandwidth,the AFSA intelligent optimization is used to design the optimum bandwidth of MIMU-aided PLL,which avoids the complex derivation process of the tracking error function versus bandwith.According to the performance of MIMU-aided PLL,the optimum bandwidth and threshold bandwidth of third order PLL under dynamic conditions are used to make assessment in this study,the results of which indicate that the MIMU aid can effectively reduce PLL tracking error and tracking threshold bandwidth,as well as improve the dynamic performance of tracking loop.Furthermore,from the simulation results,it can also be concluded that the optimum bandwidth makes sense only when it is solved under tracking threshold.3.As the performance of the filter used in nonlinear KF-based tracking loop can determine the tracking ability of GPS signal to some extent,a high-performance filter VBACKF based on VB theory is proposed to use as the prefilter in tracking loop.Firstly,the performance of VBACKF is verified by a classical nonlinear model often used in testing filter validity.Secondly,the validity of VBACKF-based tracking loop is evaluated by tracking the ideal GPS signal under high dynamic conditions.Besides,the influence of measurement model on tracking loop is also investigated.On the basis of the above nonlinear KF-based tracking loop,a GPS vector tracking loop is established,and the state space models of the vector center filter in pseudorange formulation and position formulation are introduced detailedly.As to the performance of GPS vector tracking loop established above,the paper introduces its superior performance via variance analysis of the pseudorange error;then makes simulation on the field experimental GPS IF signal by GNSS-SDR,the results of which indicate that the tracking performance and positioning performance of vector tracking loop outperform those of the scalar tracking loop.Furthermore,from the performance comparison of EKF+EKF-,EKF+VBACKF-and CKF+VBACKF-vector tracking loops,it can also be inferred that the improvement of filter performance can improve the performance of vector tracking loop.4.The disadvantages of the filter used in centralized DI structure include its high dimension,and it needs to work at high-frequency condition,therefore,the research of this paper focuses on cascaded DI structure.In cascaded DI system,the I/Q signals are firstly preprocessed by the filter,and then the results are used as the information sources of measurement for the master filter.In this way,the dimension of master filter is reduced,and the working frequency of filter is divided into two parts: the prefilter works at high-frequency condition,while the master filter works at low frequency.For cascaded DI system,the state space model of master filter is described in detail,which is used as a basis to introduce the observability analysis method of master filter.In order to improve the processing efficiency of cascaded DI system,the study focused on a method of reducing filter dimension,which firstly analyzes the impossibility for the master filter to reduce its dimension,and then takes the dimension reduction of prefilter as research focus.After deeply analyzing the relation of the state variables in traditional prefilter,the closed carrier-aided code loop tracking is removed,and the state variables are reduced to carrier phase error,carrier frequency error and carrier frequency rate error.Meanwhile,the dimension of measurement in prefilter is also reduced based on the above three-dimensional state variables.For the simplified cascaded DI structure model proposed above,the state space model of the prefilter and the control mode of carrier/code NCO are given in detail in the paper.After simulation comparison between the simplified cascaded DI system and the unsimplified cascaded DI system via MATLAB,the results show that the navigation performance of the simplified cascaded DI system and unsimplified cascaded DI system is basically the same.