Total least squares (TLS) parameter estimation is an alternative to least squares (LS) estimation when there are errors in both data matrix and observation vector. Especially, when some of the columns, not all, of the data matrix A are free of error, we call it a mixed LS-TLS problem. Accordingly, a sequential algorithm for solving a mixed LS-TLS problem is proposed here. The proposed algorithm employs an efficient algorithm to locate the minimum eigenpair, instead of singular value decomposition (SVD) which is computationally exacting. The proposed algorithm is applied to an accelerometer model to identify error parameters which are very important in inertial navigation systems (INS).
An LS-TLS-based sequential identification algorithm: Application to an accelerometer
IEEE Transactions on Aerospace and Electronic Systems ; 37 , 2 ; 607-618
2001-04-01
374219 byte
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
PAPERS - An LS-TLS-Based Sequential Identification Algorithm: Application to an Accelerometer
Online Contents | 2001
|MEMS Accelerometer Design Optimization Using Genetic Algorithm
Tema Archiv | 2013
|Axle Box Accelerometer Signal Identification and Modelling
Springer Verlag | 2020
|Springer Verlag | 2023
|Model Set Identification of Accelerometer with Nonparametric Uncertainty
British Library Conference Proceedings | 1994
|