Applications of state estimation in aircraft flight-data analysis

Bach, R. E.; Wingrove, R. C.

doi:10.2514/3.45164

Cited by 35 publications

(7 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2. Since the external forces include both aerodynamic and propulsive loads, propulsive forces (engine thrust) are removed from the estimated loads, which can be determined from monitored engine parameters, which is a common approach for flight data analysis 30 . 3.…”

Section: E Generalized Aerodynamic Performance Degradation Detectionmentioning

confidence: 99%

Aircraft Flight Envelope Determination Using Upset Detection and Physical Modeling Methods

Keller

McKillip

Kim

2009

AIAA Guidance, Navigation, and Control Conference

View full text Add to dashboard Cite

The development of flight control systems to enhance aircraft safety during periods of vehicle impairment or degraded operations has been the focus of extensive work in recent years. Conditions adversely affecting aircraft flight operations and safety may result from a number of causes, including environmental disturbances, degraded flight operations, and aerodynamic upsets. To enhance the effectiveness of adaptive and envelope limiting controls systems, it is desirable to examine methods for identifying the occurrence of anomalous conditions and for assessing the impact of these conditions on the aircraft operational limits. This paper describes initial work performed toward this end, examining the use of fault detection methods applied to the aircraft for aerodynamic performance degradation identification and model-based methods for envelope prediction. Results are presented in which a model-based fault detection filter is applied to the identification of aircraft control surface and stall departure failures/upsets. This application is supported by a distributed loading aerodynamics formulation for the flight dynamics system reference model. Extensions for estimating the flight envelope due to generalized aerodynamic performance degradation are also described.

show abstract

Section: E Generalized Aerodynamic Performance Degradation Detectionmentioning

confidence: 99%

Aircraft Flight Envelope Determination Using Upset Detection and Physical Modeling Methods

Keller

McKillip

Kim

2009

AIAA Guidance, Navigation, and Control Conference

View full text Add to dashboard Cite

show abstract

“…Using this ZOH approximation, the integral and thus the gradients can be evaluated. These gradients are then used to update the parameters 6 by a quasi-Newton procedure: (9) to minimize /. The inverse of the Hessian matrix (Jee)~l, which is difficult to compute, will be estimated using the ranktwo update procedure 14 ' 15 presented in Appendix B. Smoothing with the new set of parameters, 0 new , is performed again, followed by a parameter update.…”

Section: Identification Algorithm For Nonlinear Systemsmentioning

confidence: 99%

“…(Al) and (A2); the performance measure J, Eq. (A3); and the gradient matrices /,(/), / w (0, and h x (i) 9 Eqs. (A6).…”

Section: Nonlinear Backward Information Filter Forward Smoothermentioning

confidence: 99%

“…f c [x(t)9 u(t)Mt),t,0]dt(A20) and then compute and store the performance measure J by numerically integrating• = '/2 [x(t 0 )-x 0 ] T P 0 -' [x(t 0 )-x 0 ]compute and store the discrete gradient matrices f x (i),f w (i), and h x (i) by linearizing the continuous nonlinear dynamics equations along the nominal trajectory and discretizing them:f x (i) = DISC f w (i) = DISC h x (i) -DISC dx df c (ti),u(ti)Mti),tJ dx \x(ti),u(ti)iti (A22)…”

mentioning

confidence: 99%

See 1 more Smart Citation

Nonlinear smoothing identification algorithm with application to data consistency checks

Idan

1993

Journal of Guidance, Control, and Dynamics

View full text Add to dashboard Cite

A parameter identification algorithm for nonlinear systems is presented. It is based on smoothing test data with successively improved sets of model parameters. The smoothing, which is iterative, provides all of the information needed to compute the gradients of the smoothing performance measure with respect to the parameters. The parameters are updated using a quasi-Newton procedure, until convergence is achieved. The advantage of this algorithm over standard maximum likelihood identification algorithms is the computational savings in calculating the gradient. This algorithm was used for flight-test data consistency checks based on a nonlinear model of aircraft kinematics. Measurement biases and scale factors were identified. The advantages of the presented algorithm and model are discussed. Nomenclature a x ,a y ,a z = body axis components of the translational acceleration ax cg> a ycg> a zc = b°dy ax * s components of the translational acceleration at the center of gravity b = measurement bias vector, dimension m b% -measurement bias d(.) = process noise input in the SMACK model E = m x m diagonal matrix of measurement scale factors f c = nonlinear continuous dynamics vector function, dimension n fd = nonlinear discrete dynamics vector function, dimension n fxtfw -gradient matrices of the vector function f c h -altitude h m -nonlinear measurement vector function, dimension m h x -gradient matrix of the vector function h m J= performance measure L = transformation matrix from Earth to aircraft body coordinate system L p ,L da = aircraft stability and control derivativeŝ ,f^,f^ = sensor location measured from the center of gravity in the aircraft body coordinates n% = measurement noise P 0 = initial conditions weighting matrix p,q,r = body axis components of the angular rate Q = process noise weighting matrix R = output error weighting matrix (R, (B,8 = ground track data: slant range, bearing, and elevation angles s^ = measurement scale factor T = transformation matrix from aircraft body to Earth coordinate system ti = time at the sampling instants, such that £/ = t Q + iAt for / = 1,..., TV and At is the time between samples u -known input vector, dimension c u a> v a>w a = body axis components of the translational velocity at the center of gravity assuming calm atmosphere

show abstract

“…The steady-state Kalman gain matrix # is given by (6c) (6d) (7) where Q is the steady-state covariance of the predicted state, given as the limit of the discrete-time Riccati equation (8) If all unstable modes of the system are (^^-controllable an4 04,C)-observable, then Eq. (8) is guaranteed to converge to a unique steady-state solution independent of the initial covariance.…”

Section: Maximum Likelihood Estimatormentioning

confidence: 99%