Nonlinear Initial Alignment In Inertial Frame For SINS

The thesis has investigated in the nonlinear initial alignment of strapdown inertial navigation system(SINS).The contents involve the initial alignment method based-on multiple vector observations in inertial frame,modeling of the nonlinear initial alignment in inertial frame and the corresponding filtering technology,the transfer alignment in inertial frame and the in-motion alignment aided with GPS.The main research contents include:(1)Optimization-based self-alignment algorithm using multiple vector observations in inertial reference frame is investigated.The time-vary direct-cosine matrix is decomposed into three new attitude matrices: the first one describing the inertial angular motion of the body frame can be updated by integrating gyros measurements;the second one built of the geography and the earth rotation message can be analytically solved with already known parameters;the third one is actually the attitude matrix between the body frame and the navigation frame at the start time of the alignment,which can be estimated based on the multiple vector observations constructed by the slowly changed special force in the inertial frame.The common attitude estimation methods using multiple vector observations are introduced in detail,and a simplified algorithm based on Rodrigues parameters is deduced rigorously.(2)Two simplified models for the estimation of the third attitude matrix under large misalignment angles are established in inertial frame,based on quaternion description and Rodrigues description respectively.The common states of the two models include the special force integration error in inertial frame,the calculation error of the attitude between body frame and inertial frame,and the inertial sensor errors;the different states are the quaternion and the Rodrigues parameters,which are used to equally describe the third attitude matrix respectively.The system equations of the two models based on these states are all linear,and measurement equations are only second order polynomials of the states,which have weak nonlinear property.(3)An efficient filtering algorithm for the special model which has linear system equations and second order nonlinear measurement equations,is designed.The standard Kalman filter is used to do the time update,and the second order extended Kalman filter is included to finish the measurement update.Comparing with the standard Kalman filter,the new filtering algorithm only adds easy modifications of the innovation and innovation covariance.Further more,when applied to the quaternion model,the optimization form of the new filtering method is deduced under the quaternion norm constraint;when applied to the Rodrigues model,the singular problem encountered is discussed and solved.(4)The application of the afore designed filtering algorithm in transfer alignment(TA)is discussed.Large attitude between the master IMU and slave IMU may cause hard-to-overcome difficulties for the traditional transfer alignment method.Aiming to overcome these difficulties,a new transfer alignment algorithm is designed in inertial frame,and four feasible matching vectors are introduced,i.e.,special force matching,special force integration matching,angular rate matching,angular rate integration matching.A detailed implementation of the new TA algorithm using special force integration plus angular rate integration matching is given,and the simulation under the Wing-rock maneuver is performed to validate the new algorithm.(5)An in-motion alignment algorithm of SINS/GPS based on the inertial filtering method is investigated.First,the implementation of the in-motion alignment method is deduced in detail.Second,an accurate velocity calculation method using incremental carrier-phase observable of GPS is introduced.Then,the observability of the nonlinear in-motion alignment is analyzed and an observable matrix is obtained.At last,experiments are performed using MEMS IMU/GPS on land vehicle to validate the algorithm.