On-line Alignment And Calibration Technique Of Fiber Optic Gyroscope SINS

In order to meet the requirements of the current marine navigation system, the researchand engineering of fiber optic gyro Strapdown Inertial Navigation System (SINS) will befaced with a critical and urgent task. Improving the precision of navigation systems based oninherent hardware is one of the urgent problems needed to be resolved in the present inertialtechnology field. Based on the actual demands of improving marine SINS’s output accuracy,this dissertation investigates two key technologies involved in SINS including on-line initialalignment and calibration. The content mainly comprises of the following five aspects:1、Based on the reference vectors of coarse alignment selected in the inertial frame, thisdissertation studies the error analysis of the inertial mooring alignment. The analyticalexpressions relating gyro drift and accelerometer bias to misalignments are derived.Accuracy comparison with traditional algorithms is performed, which can provide theoreticalanalysis basis for inertial alignment. Through analyzing the effect of the two velocityreference vectors sampling points on the alignment accuracy, the optimal velocity referencevectors sampling points are derived.2、With the deficiency of inertial alignment for filtering gyro random noise, an improvedon-line alignment based on Hidden Markov Model Kalman Filter (HMMKF) is presented,which uses the real-time filtering characteristic of HMMKF. HMMKF filtering equations aresimplified and the filtering mechanism is discussed. Comparisons with Digital Filter (DF)prove that the real-time performance of HMMKF can guarantee the requirement of initialalignment. The relationship between the misalignments and inertial instrument calibrationerrors is illustrated. To make misalignments fully reflect inertial instrument calibration errors,a fast six-position calibration algorithm based on HMMKF is proposed and navigationexperiment verifies its validity.3、An analytic method used to analyze the observability of initial alignment is researched.On the study basis of the relationship between alignment accuracy and yaw, pitch and roll,the optimal attitude involved in alignment is derived. And then, based on the requirements offurther compensation for residual misalignments, a two time fast autonomous alignmentmethod with velocity errors and residual misalignments as observables is performed, whichuses the residual misalignments expanded to state variables. The system model and theabservation model introduced new state variables are derived, and the simulation is carriedout too.4、The different projective components of inertial instrument errors on the geographic coordinate frame with Instrument Measurement Unit (IMU) different positions are discussed.According to the accurate transposition control of the rotating mechanism, a three-positionalignment algorithm for automatically eliminating the misalignments caused by unobservablestates is proposed. The optimal three positions of IMU are derived. Comparisons with fixed-position alignment and two-position alignment prove that three-position alignment algorithmcan effectively improve alignment accuracy. On the research of the relationship betweenmulti-position misalignments and inertial instrument errors, a three-position on-fieldcalibration method based on uniaxial rotation SINS is proposed. The three-positioncalibration equations to estimate the inertial instrument biases are derived, and the optimalthree positions are verified both by theoretical analysis and by experiments.5、Turntable error model building based on the rotation quaternion is studied. Theinfluence of turntable tilt errors, turntable angular position errors and non-orthogonality onthe six-position calibration precision of inertial measurement units is discussed, and therelationship between turntable errors and inertial instrument calibration accuracy is alsoderived. Based on three-axis turntable providing more rotation positions, a ten-positioncalibration algorithm to calculate accelerometer biases and installation angles with horizontalmisalignments is proposed. Through the separation of calibration parameters and analysis onthe influence for turntable errors to calibration accuracy, the optimal ten-positionaccelerometer calibration equations based on kalman filter are derived. On the basis ofsix-position calibration, Precision estimation of accelerometer residual calibration errors isachieved and the accelerometer calibration accuracy is improved.