| A new approach, which enhances Inertial Navigation System (INS) accuracy by modernizing the INS algorithmic part rather than improving the inertial sensors, is investigated. This dissertation proposes block-processing development of the INS algorithmic part in the frequency domain to improve the INS accuracy performance. Considering current demands for inertial sensor miniaturization and cost decrease, and operation of the INS as a part of integrated Global Positioning System/INS systems, the block-processing approach is applied to improve short-term accuracy characteristics of the strapdown, low-grade INS. A generic frequency-domain INS computation starts with the reconstruction of continuous-time signals from blocks of input discrete samples by implementing the Fourier Transform (FT) compensated for boundary discontinuities. Next, analytical transformation of FT spectrums of reconstructed input signals into spectrums of navigation outputs is carried out in the frequency domain. The generic frequency-domain computation is utilized to develop frequency-domain attitude determination, coordinate transformation, integration, and INS calibration procedures. Test results presented verify that the frequency-domain approach shows a significant improvement over conventional, discrete-time methods in the ability to reconstruct the asculling and coning motion and subsequent reduction in sculling and commutation errors. The error reduction demonstrated is mostly critical when frequencies of oscillatory motion components are comparable to half the sampling frequency of input inertial measurements. An order of magnitude reduction in the calibration methodical error is demonstrated by implementing frequency-domain techniques for the calibration of the strapdown, low-grade INS. In addition, frequency-domain development of the INS calibration procedure improves observability of calibrated parameters and, as a result, compensates for the unbounded growth of degradation of inertial accuracy in cases of singularity, contrary to conventional calibration approaches. |
