A Fault-Tolerant Estimator for Redundant Systems

Broen, R. B.

doi:10.1109/taes.1975.308185

Cited by 6 publications

(1 citation statement)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The FDI method: Wilcox proposed a nonlinear or adaptive filter using maximum likelihood theory for detecting the failure of an RSINS with the dodecahedron configuration containing six gyroscopes and six accelerometers [9]. Broen introduced a non-stationary weighting matrix associated with the least squares estimator for the nonlinear failure detection and isolation feature of the RSINS [10]. Kim et al proposed the extended parity space approach which was suitable for fault detection and isolation of the system with one faulty sensor or two faulty sensors [11].…”

Section: Introductionmentioning

confidence: 99%

Self-alignment of full skewed RSINS: Observability analysis and full-observable Kalman filter

Song

Zhang

2014

J. of Syst. Eng. Electron.

View full text Add to dashboard Cite

Traditional orthogonal strapdown inertial navigation system (SINS) cannot achieve satisfactory self-alignment accuracy in the stationary base: taking more than 5 minutes and all the inertial sensors biases cannot get full observability except the up-axis accelerometer. However, the full skewed redundant SINS (RSINS) can not only enhance the reliability of the system, but also improve the accuracy of the system, such as the initial alignment. Firstly, the observability of the system state includes attitude errors and all the inertial sensors biases are analyzed with the global perspective method: any three gyroscopes and three accelerometers can be assembled into an independent subordinate SINS (sub-SINS); the system state can be uniquely confirmed by the coupling connections of all the sub-SINSs; the attitude errors and random constant biases of all the inertial sensors are observable. However, the random noises of the inertial sensors are not taken into account in the above analyzing process. Secondly, the full-observable Kalman filter which can be applied to the actual RSINS containing random noises is established; the system state includes the position, velocity, attitude errors of all the sub-SINSs and the random constant biases of the redundant inertial sensors. At last, the initial selfalignment process of a typical four-redundancy full skewed RSINS is simulated: the horizontal attitudes (pitch, roll) errors and yaw error can be exactly evaluated within 80 s and 100 s respectively, while the random constant biases of gyroscopes and accelerometers can be precisely evaluated within 120 s. For the full skewed RSINS, the self-alignment accuracy is greatly improved, meanwhile the self-alignment time is widely shortened.

show abstract

Section: Introductionmentioning

confidence: 99%