| At present，the marine inertial navigation system with mechanically dithered ringlaser gyroscope as the core inertial sensors has been widely used in those militarydeveloped countries led by the U.S., and it plays an important role in the field ofhigh-precision navigation of surface ships and submarines. Rotating modulationtechnique has been universally adopted in long-endurance, high-precision marineinertial navigation systems. For the single axis rotation modulation INS, errors along therotation axis can not be modulated, which limit the system accuracy. However, thedouble-axis rotation modulation can modulate the apparatus error in three directions;therefore, it inhibits the time accumulated navigation errors. Based on some keytechnologies of double-axis rotation modulation INS with mechanically dithered RLG isdiscussed, aiming to enhance the long-endurance accuracy of the RLG INS with somedegrees and solve the bottleneck problem of Chinese Navy heading for the ‘dark blue’.The research work includes the following five aspects:1．The design and analysis of the optimization scheme of double-axis rotationmodulation INS. Based on the SINS error equations, the modulation mechanism of gyrobias, scale factor error, installation error is analyzed in detail, and judgment rules of theerror modulation effects are put forward. On the comparison of the modulation effectsof the sixteen-sequence scheme and sixty-four-sequence scheme, an improved schemeof sixty-four-sequence is presented, which is far more reasonable and practical.2．Study on the filtering algorithm and random errors of mechanically ditheredRLG which is the core inertial sensor in double-axis rotation modulation INS. A newmethod named grey-time series modeling is proposed, which has integrated themetabolic GM(1,1) model and Time series model, and completed the online AR(2)modeling of drift data, then the random drift of gyro’s output is inhibited effectively byKalman filter. A filtering method of RBF neural network which combines HarmonicWavelet pre-filtering is proposed. This kind of filtering method using trained RBFneural network improves the accuracy of mechanically dithered RLG. The evaluation ofthe performance of high-accuracy mechanically dithered RLG is studied, the demerit ofthe Allan variance have been pointed out, and a new method of analyzing the collectingdata of high-accuracy mechanically dithered RLG using Dynamic Allan variance(DAVAR) has been proposed.3．Study on system-level temperature compensation technology of INS withmechanically dithered RLG. Combined with the practical application environment ofthe marine INS, a reasonable experiment of system temperature compensation has beendesigned. The effects of temperature compensation to inertial sensors adopting multiple linear regression method and robust regression method are compared. Wheninterference appears, the compensation accuracy using robust regression method issignificantly better than that adopting multiple linear regression method. The feasibilityof LS-SVM method for temperature compensation of mechanically dithered RLG isdiscussed. Experiments indicate that the LS-SVM has a optimazed ability of nonlinearfitting, so it is useful for resolving the problem that the limited fitting ability oftraditional least squares or iterative least squares.4．Study on calibration technique of INS with mechanically dithered RLG. Thispaper clearly illustrates the mathematical principles of the error actuation and separationby detailed derivation of the expression of the difference of speed change rate beforeand after each rotation. This paper designed a30-dimensional Kalman filter to completethe calibration experiment, and achieves excellent experimental results. Throughexploring the double-axis rotation INS self-calibration technology in the marineenvironment, this paper designed a double-axis rotation self-calibration program, andpointed out that in addition to three installation errors, the rest errors all can becalibrated out. By taking advantage of the vehicle maneuver, the double-axis rotationINS also realizes full parameter calibration.5．The algorithm design and navigation experiments of double-axis system. Byanalyzing the attitude extraction algorithm, the reason why the double-axis rotationadopts rotation around the roll and heading axis, not the pitching and heading axis, hasbeen pointed out. The non-rotating static navigation experiment、single-axis rotationnavigation experiment、 double-axis rotating navigation experiment and dampingexperiment have all been done. The double-axis rotating navigation experiment using50-type RLG has achieved the accuracy of maximum error1.02nm/120hours, andthere is no divergent trends in the position errors. During off-line simulation ofhorizontal damping and full damping of the double-axis rotation system, the maximumhorizontal damping navigation positioning error is0.6nautical mile during the120hours, the maximum full damping navigation positioning error is0.56nautical mileduring the120hours. The double-axis rotating navigation experiment using90-typeRLG has overmatched the accuracy of0.6nm/14days. |
PDF Full Text Request