Transfer Alignment For Strapdown Inertial Navigation Systems

Strapdown inertial navigation system (SDINS) is a dead-reckoning navigation system of which the performance depends heavily on the accuracy of the initial alignment, determination of the transformation matrix from the body frame to the navigation frame. Moreover the estimation accuracy and convergence time of the attitude misalignment determine the performance of the whole alignment process. Transfer Alignment (TA) is the process of simultaneously initializing and calibrating at last one slave inertial navigation system (SINS) using the accurate information from one Master inertial navigation system (MINS). In general, this process is accomplished by calculating the difference of navigation solutions between SINS and MINS to form observations which are then used in a Kalman Filter (KF) to recursively estimate and correct SINS attitude and inertial sensor errors.Numerous techniques about the problem of TA existed in the literature were firstly studied. Several TA algorithms each of which process different measurement types were designed and implemented. These algorithms have been validated by simulation in a flexure and vibration environment for typical airborne applications, the advantages and disadvantages of each algorithm were analyzed and the dependence of TA performance on KF system model, vehicle maneuvers and alignment duration were investigated in detail.Especially, in order to exploit any low level natural angular motions of a vehicle to address the problem of TA under the vehicle's low dynamics, a new filter algorithm for TA was proposed by augmenting the velocity-match with angular rate-match, and an improved method using integrated velocity and integrated angular rate matching was then introduced to reduce the effect of the vehicle flexure for improving the performance of TA. The advantages of two proposed algorithms were demonstrated by simulations for the alignment and calibration of airborne sensors and transfer alignment of shipborne weapon INS. In addition, strapdown INS velocity and attitude error propagation models for large angle errors were developed to deal with the problem of TA with unknown initial conditions, and then the TA approach to handle the non-linearity problem caused by the initial attitude error uncertainty was implemented by applying KF algorithm based on singular value decomposition (SVD).Further, the Wavelet Multi- Resolution -Analysis (WMRA) technique was proposed in this thesis as an efficient pre-filter for SINS inertial sensors output. Applying this pre-filtering process aims to improve the sensors'signal-to-noise ratios, to remove those errors mixed with the vehicle motion dynamics and to provide more reliable data to the KF-based TA algorithms. In addition, the cascade denoising algorithm was proposed to overcome the limitations of existing wavelet denoising. The effectiveness of the proposed cascade denoising method in improving the performance of TA was then evaluated by a set of simulations.