Optimized Filter Design for Non-Differential GPS/IMU Integrated Navigation.
Technische Universität, Darmstadt
[Ph.D. Thesis], (2015)
Available under Only rights of use according to UrhG.
Download (5MB) | Preview
|Item Type:||Ph.D. Thesis|
|Title:||Optimized Filter Design for Non-Differential GPS/IMU Integrated Navigation|
The endeavours in improving the performance of a conventional non-differential GPS/MEMS IMU tightly-coupled navigation system through filter design, involving nonlinear filtering methods, inertial sensors' stochastic error modelling and the carrier phase implementation, are described and introduced in this thesis. The main work is summarised as follows.
Firstly, the performance evaluation of a recently developed nonlinear filtering method, the Cubature Kalman filter (CKF), is analysed based on the Taylor expansion. The theoretical analysis indicates that the nonlinear filtering method CKF shows its benefits only when implemented in a nonlinear system. Accordingly, a nonlinear attitude expression with direction cosine matrix (DCM) is introduced to tightly-coupled navigation system in order to describe the misalignment between the true and the estimated navigation frames. The simulation and experiment results show that the CKF performs better than the extended Kalman filter (EKF) in the unobservable, large misalignment and GPS outage cases when attitude errors accumulate quickly, rendering the psi-angle expression invalid and subsequently showing certain nonlinearity.
Secondly, the use of shaping filter theory to model the inertial sensors' stochastic errors in a navigation Kalman filter is also introduced. The coefficients of the inertial sensors' noises are determined from the Allan variance plot. The shaping filter transfer function is deduced from the power spectral density (PSD) of the noises for both stationary and non-stationary processes. All the coloured noises are modelled together in the navigation Kalman filter according to equivalence theory. The coasting performance shows that the shaping filter based modelling method has a similar and even smaller maximum position drift than the conventional 1st-order Markovian process modelling method during GPS outages, thus indicating its effectiveness.
Thirdly, according to the methods of dealing with carrier phase ambiguities, tightly-coupled navigation systems with time differenced carrier phase (TDCP) and total carrier phase (TCP) as Kalman filter measurements are deduced. The simulation and experiment results show that the TDCP can improve the velocity estimation accuracy and smooth trajectories, but position accuracy can only achieve the single point positioning (SPP) level if the TDCP is augmented with the pseudo-range, while the TCP based method's position accuracy can reach the sub-meter level. In order to further improve the position accuracy of the TDCP based method, a particle filter (PF) with modified TDCP observation is implemented in the TDCP/IMU tightly-coupled navigation system. The modified TDCP is defined as the carrier phase difference between the reference and observation epochs. The absolute position accuracy is determined by the reference position accuracy. If the reference position is taken from DGPS, the absolute position accuracy can reach the sub-meter level.
For TCP/IMU tightly-coupled navigation systems, because the implementation of TCP in the navigation Kalman filter introduces additional states to the state vector, a hybrid CKF+EKF filtering method with the CKF estimating nonlinear states and the EKF estimating linear states, is proposed to maintain the CKF's benefits while reducing the computational load. The navigation results indicate the effectiveness of the method.
After applying the improvements, the performance of a non-differential GPS/MEMS IMU tightly-coupled navigation system can be greatly improved.
|Place of Publication:||Darmstadt|
|Classification DDC:||000 Allgemeines, Informatik, Informationswissenschaft > 004 Informatik
600 Technik, Medizin, angewandte Wissenschaften > 600 Technik
|Divisions:||13 Fachbereich Bau- und Umweltingenieurwissenschaften
13 Faculty of Civil and Environmental Engineering > Institute of Geodesy
13 Faculty of Civil and Environmental Engineering > Institute of Geodesy > Physical and Satellite Geodesy
|Date Deposited:||11 Sep 2015 07:04|
|Last Modified:||11 Sep 2015 07:04|
|Referees:||Becker, Prof. Matthias and Eichhorn, Prof. Andreas|
|Refereed:||7 July 2015|