Anda di halaman 1dari 34

Measurements for Flight Vehicle System Identification

Measurements
Maneuver
Input

Optimized
Input

Actual
Response

FlightVehicle

Data Collection
& Compatibility

Methods
APrioriValues,
lower/upper
bounds

Estimation
Algorithm/
Optimization

Models
Model
Structure

Mathematical
Model/
Simulation

Identification
Criteria

Response
Error

Parameter
Adjustments
ModelResponse

Identification Phase
Complementary
Model
Flight Data
Validation
Dr. Ravindra Jategaonkar

Validation Phase

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/1

This page is left intentionally blank.

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/2

General Aspects
Parameters
Control surface deflections
- Potentiometers, LVDT

Linear accelerometers

Data Consistency
Calibration errors
Flight path reconstruction

- Accelerometers, INS

Angular rates
- Rate Gyros, INS
strap-down platform

Attitude angles
- INS

Engine parameters

Current trends
Flush airdata sensing system
- multi-hole probes

GPS/DGPS
- position, attitude, wind

Pilot forces
- force transducer

Air data
- flow angles - vanes
- Airspeed
- Pressure sensors

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/3

Kinematic Consistency Checking


General Approach, sensor model and estimation algorithm
- Toensurethatthemeasurementsare
consistentanderrorfree.
- Inertialmeasurements(accelerationsand
angularratesarehighlyaccurate.
- Kinematic equationswithnouncertainties;
Accurateinformationaboutaircraftstate;
- Meanstocalibrateparameterswithlower
accuracy(angleofattack&sideslip).
Kinematic Equations
LinearAccelerations
Rotationalrates
ax,a y,az,p,q,r

Velocitycomponents
(u,v,w)
Attitudeangles

Sensor calibration model


Scale factor and bias

pd m = K
pd m = K

pdyn nb + pd
pdyn nb + pd

Time delay

pd m( t ) = K pdyn ( t q )nb( t ) + pd

Parameter estimation
- Output error method for nonlinear systems
- Bounded-Variable Gauss-Newton algorithm
- Multiple experiment evaluation

Initialconditions:u 0,v0,w0,p0,q0,r0
Scalefactors:K p,Kq,K r
Bias: ax,ay,az,p,q,r
Dr. Ravindra Jategaonkar

K , pd , K , pd , , , q

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/4

Data Compatibility Check - Equations (1)


Kinematic Equations:
u& = qw + rv g sin + a x

v& = ru + pw + g cos sin + a y


& = pv + qu + g cos sin + a y
w
& = p + q sin tan + r cos tan
& = q cos r sin
& = q sin sec + r cos sec
h& = u sin v cos sin w cos sin

u ( 0) = u 0
v ( 0) = v 0
w ( 0) = w 0

(0) = 0
( 0) = 0
( 0) = 0
h ( 0) = h 0

Kinematic equations are exact in terms of motion variables:


(u, v , w) ( , , ) ( x g , y g , h ) ( p, q, r ) ( a x , a y , a z )
No modeling errors (as in the case of aerodynamic models).

- First three triples are state variables.


- Assuming measurements of last two triples, kinematic equations
provide complete information for state estimation (FPR).
- Initial conditions are estimated as unknowns
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/5

Data Compatibility Check - Equations (2)


Introduce measurement bias errors in linear accelerations
and angular rates:
State Equations:
u&

= ( q q) w + ( r r )v g sin + ( a x a x )

u (0) = u0

v&

= ( r r )u + ( p p ) w + g cos sin + ( a y a y )

v (0) = v0

w& = ( p p )v + ( q q)u + g cos sin + ( a y a y )

w(0) = w0

&

= ( p p ) + ( q q ) sin tan + ( r r ) cos tan

( 0) = 0

&

= ( q q) cos ( r r ) sin

( 0) = 0

&

= ( q q) sin sec + ( r r ) cos sec

( 0) = 0

h&

= u sin v cos sin w cos sin

h (0) = h0

Input variables: Linearaccelerations(referredtoCG): a x , a y , a z


p , q, r
Angularrates
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/6

Data Compatibility Check - Equations (3)


Introduce sensor models: Scale factor and bias corrections
(Simplified models)
Observation Equations :

AoA and AoS (Vane)

AoA and AoS (5-Hole probe)

m
m

pd ,m

Vm

m
m
m
hm

=
=
=
=
=

K local +
K local +
V
K +
K +

pd ,m
pdyn,m

= K pdyn local + pd
= K pdyn local + pd
2
= 12 Vlocal

= K +
= h + K h local + K h local

Accountfortimedelays:e.g. m = K (t ) +
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/7

Flight Path Reconstruction Techniques (1)


Two types of algorithms:
1) Extended Kalman Filter
2) Output Error Method
depending upon the assumptions we make regarding the noise in
the input variables ( a x , a y , a z , p, q, r, p& , q& , r&)
Deterministic Approach
Accounts for measurement noise in output variables only.
The state estimation problem reduces to simple numerical integration.
x& (t ) = f [ x (t ), u (t ), ] ,

x (t 0 ) = x 0

y (t ) = g[ x (t ), ]
z (t k ) = y (t k ) + v (t k )

x = [ u v w h ]T
u = [a x a y a z p q r ]T
y = [V h ]T

The unknown parameters are:


= [ a x a y a z p q r K K , ]T

x 0 = [ u 0 v 0 w 0 0 0 0 h 0 ]T

Apply output error method


Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/8

Flight Path Reconstruction Techniques (2)


Stochastic Approach
Accounts of measurement noise in both input and output variables.
x& (t ) = f [ x (t ), u m (t ) w(t ), ],
y (t ) = g[ x (t ), ]
z (t k ) = y (t k ) + v (t k )

x (t 0 ) = x 0

w(t ) noise in the measured inputs


u = [a x a y a z p q r ]T

v(t k ) noise in the observation variables


y = [V h ]T

Note:
The noise w(t ) in this context is not the same as the atmospheric turbulence
Presence of input noise makes system stochastic:
Simple numerical integration not possible (due to nonmeasurable input noise).
Needs state estimator (Extended Kalman filter)
EKF will be addressed later
Another approach: EBM (Estimation-Before-Modeling)
For details see Ref. 1.
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/9

FPR Identifiability of Sensor model


Scale factor and bias may be correlated:
Due to differences in order of
magnitude between the trim
(nominal) value and bias error.
Small variations in scale factor
(usually one) multiplied by the
large trim value (e.g. V or )
appearing on the right hand
sides) can compensate for small
zero shifts without affecting
the match significantly.

20
deg
0
-5
300
kt

V
100
0.3
g
ax
0
-0.1
0
g
az
-2
0

400

time

800

1200

1600

To avoid correlation:
1) Combination of dynamic maneuvers are different trim points and of
dynamic and quasi-steady maneuvers
2) Reformulate the observation equation: m = K ( 0 ) + 0 +
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/10

Consistency Checking of Control Inputs


Kinematic consistency checking provides an a-posteriori means to verify
recorded data and to estimate instrument errors pertaining to the aircraft
motion variables.
It is an off-line procedure which can be applied after flight testing.
Besides these motion variables, aerodynamic model identification also needs
the measurements of control surface deflections.
The errors of the measurements of control surface deflections!!!
Unfortunately, there is no way to check the accuracy of these measurements,
because there are no kinematic relations between them and other variables.
This is a fundamental problem, for which there is no easy solution.
Hence, calibration of these independent inputs needs great care and
considerable effort is spent on laboratory and in-situ calibration of these
measurements, and on re-verification at regular intervals.
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/11

Data Compatibility Check: Examples (1)


Example 1: Demonstration of the procedure
Three dynamic maneuvers:
- an elevator input maneuver exciting the short period motion (25 s),
- an aileron input maneuver resulting in bank-to-bank motion (30 s)
- a rudder input maneuver leading to Dutch roll motion (30 s).

The flight maneuvers under seemingly steady atmospheric conditions,


- turbulence is considered negligible.
States: x = [u, v, w, , , , h]

Outputs: y = [V , NB , NB , , , , h]

Inputs: u = [a x , a y , a z , p, q, r , p& , q& , r&]


Unknown parameters: = [ a x a y a z p q r ]T

Apply Output error method.


To start with all parameters set to zero.

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/12

Data Compatibility Check: Examples (2)


Example 1: Demonstration of the procedure
2
ax
(m/s 2)

Input variables:

ay
(m/s 2)

-2
5
0

az
(m/s 2)

-5
0
-10

p
(/s)

-20
50
0

0
-10
10

r
(/s)

Run the test case:


/FVSysID/chapter04/
test_case = 22
niter_max = 0

q
(/s)

-50
10

0
-10

Dr. Ravindra Jategaonkar

10

20

30

40
50
Time in sec

60

70

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

80

90

Measurements/13

Data Compatibility Check: Examples (3)


Example 1: Demonstration of the procedure
V
(m/s)

140
120

()

Match for iter=0

100
10

()

-10
5
0

()

-5
50
0

()

-50
10
0

()

-10

20
10
4
0 x 10

h
(ft)

1.7
1.6
1.5

Dr. Ravindra Jategaonkar

10

20

30

40
50
Time in sec

60

70

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

80

90

Measurements/14

Data Compatibility Check: Examples (4)


Example 1: Demonstration of the procedure
V
(m/s)

135
130

()

125
6

()

Run the test case:


/FVSysID/chapter04/
test_case = 22
niter_max = 10

0
-3

5
0
-5

()

40
0

()

()

-40
10
0
-5
20
10

h
(ft)

0 x 10
1.65
1.6
1.55

Dr. Ravindra Jategaonkar

10

20

30

40
50
Time in sec

60

70

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

80

90

Measurements/15

Data Compatibility Check: Examples (5)


Example 2: Rotational rates and attitude angles (special case)
Kinematic Equation
Rotationalrates

p,q,r

Attitudeangles

0.4
rad
0
-0.6
0

Initialconditions: p0,q0,r0
Scalefactors:Kp,Kq,Kr
Bias:p,q,r

Bankangle

20

+ve error

40

sec

Rotational rates and bank angle compatible


0.6
rad

Nonlinear scale factor


errors in pitch attitude

time

Pitch attitude

0
-0.4
0

20

0.6
rad

time

40

sec

40

sec

Pitch attitude

-ve error

Dr. Ravindra Jategaonkar

-0.4
0

20

time

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/16

Data Compatibility Check: Examples (6)


Reconstruction unreliable due to lack of
40
m/s
u

Kinematic Equations
LinearAccelerations
Rotationalrates
ax,ay,az,p,q,r

Velocitycomponents
(u,v,w)
Attitudeangles

20
10
m/s
v
0
-5
8
m/s
w 0

Initialconditions:u0,v0,w0,p0,q0,r0
Scalefactors:Kp,Kq,Kr
Bias:ax,ay,az,p,q,r

-8
0

Trueairspeed- measured

40

60
time

80

100

sec

100

sec

Engineering judgement

45
m/s

20
0

20

40
m/s
u
20

40

60
time

80

100

sec

20
10
m/s
v
0

Measurement of and necessary for


reliable results
Identified models subject to uncertainties
Dr. Ravindra Jategaonkar

-5
8
m/s
w 0
-8

20

40

time

60

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

80

Measurements/17

Data Compatibility Check: Examples (7)


Location of Flow Sensors: ATD VFW-614 (1)

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/18

Data Compatibility Check: Examples (8)


Location of Flow Sensors: ATD VFW-614 (2)
15
deg

Maneuver1

Maneuver2

Maneuver3

Angles of Attack

AoA1
0
13
deg

AoA2
-2
20
deg

AoA3
5
100

200

Time

300

400

Flightmeasured
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Estimated
Measurements/19

Data Compatibility Check: Examples (9)


Location of Flow Sensors: ATD VFW-614 (3)
Cross Plot of Measured Angles-of-Attack
30

AOA3 Versus AOA1

AoA3
AoA2

20

AoA1
AOA1VersusAOA1
10

AOA2VersusAOA1
0

-5
-5

10

15

20

deg

25

AoA1
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/20

Data Compatibility Check: Examples (10)


Location of Flow Sensors: ATD VFW-614 (4)
Flow Interference Effects due to Pitot Tube Wakes
14.5 to17.5 to21 bandfrom Pitot #2

14.5 to17.5 to21 bandfrom Pitot #3

PositionofPitot #2
for Flight No.V3320

PositionofPitot #2
for Flight Nos.V3321,
3322and3323

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/21

Influence of Position on Angle-of-Attack


Measured angle of attack may contain
components due to angle of sideslip

10

Angle
of
Sideslip

deg

0
-10
30

Angle
of
attack

aoa1
+ F1N min( , 0) + F1P max( , 0)

Neglecting the influence due to AoS


m _1 = K1 tan1 uaoa1 + 1
w

Dr. Ravindra Jategaonkar

aoa1

0
-10
2.4

Asymmetric influence due to AoS


m _1 = K1 tan1 uaoa1 + 1

deg

Prediction deg
error on
0
AoA
-2.4
0

200

Time

400

600

400

600

2.4

Prediction
error on
AoA

deg

0
-2.4
0

200

Time

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/22

Calibration of Angle of Attack: ATTAS VFW-614


Calibration of Angle of Attack

Flow Angle Sensors

Flight Log
4

deg

Flight log
2

Flap

FLog =1.153+FLog
0

IN(-6)
SP(1)
5
14
35

-1
120

160

200

240

kt

4
deg

RMP =1.407+RMP
2

Flap

Vane

IN(-6)
SP(1)
5
14
35

0
-1
120

Vane
200

160

240 kt

CAS

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/23

Calibration of 5 Hole Probe


Rosemount Flow Angle Sensor
P2
P

Angleof attack:
Pt
P Angleof sideslip:
2

Flow sensors

P P
= 1 2
qc K
P P2
= 1
qc K

Flight log

Dynamicpressure: qc = Pt Ps

Linear sensitivity per degree


0.08

0.06

5 Hole Probe

K
0.04

0.02
0

0.4 0.6

1.0

2.0

Mach number
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/24

Calibration of static pressure ports


Position Error
Altitude and airspeed derived from
static and total pressure
Measurement in free stream air
Distorted flow field around an a/c
Pitot static system:
- location of pressure ports
- speed
- altitude
Commonly adopted procedures to
calibrate static pressure ports:
- Trailing cone
- Tower Fly-By
Dr. Ravindra Jategaonkar

Static pressure field varies along


the length of aircraft

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/25

Calibration of static pressure ports


Determination of Position Error: Tower Fly-By procedure
Simple and accurate procedure
- fly the a/c past an observation tower
- record geometrical position, camera
- onboard data: static pressure, AGL...
- Ambient conditions, QNH

330m

Tower;
Camera

Flight tests with ATTAS


- Test matrix: 16 fly-bys, 4 flap positions
- speed range from 120 to 240 KIAS

109m

NN(Sealevel)

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/26

Calibration of static pressure ports


Determination of Position Error: Tower Fly-By results
Altitude determination

Static pressure

- simple geometrical procedure


- two independent cameras
- Radio altimeter

- computational procedure (h, P0, T0)


- Pstat time records from two sources

Flight tests results:

Flight tests results:


- Fairly good agreement
- minor deviations
variations in flight path above runway
different time points of exposure

- Good agreement between derived


and measured basis static pressure.
- Discrepancies in static pressure
measured by 5-Hole probe.
1004

1,SP

-6,IN

120-190Kt

140-240Kt

5 Flaps

145-240Kt

Altitude,ft

160

140

120

1,SP
140-240Kt

-6,IN

5 Flaps

145-240Kt

130-210KIAS

Aircraftbasis
Radioaltitude
5Holeprobe

130-210KIAS

Staticpressure,hPa

180

14

14
120-190Kt

1000

996

Altitude,Camera1
Altitude,Camera2
Radioaltimeter

100
0

Towerfly-by,Dr. Ravindra Jategaonkar

12

16

992

12

16

Towerfly-by,-

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/27

Calibration of static pressure ports


Position Error at Noseboom
- difference in 5-hole probe measured static pressure and that derived from fly-bys
- function of airspeed (covers four flap configuration, speed from 120 to 240 KIAS)
- ATTAS noseboom 1.35 times fuselage diameter
- empirical rule: noseboom length 2.5 to 3 time fuselage diameter

Position error,(Pa)

800

600

400

200
100

150

250

kts

250

Trueairspeed
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/28

Estimation of Time Delays


5000
PA

PDALFA
0
3000
PA

PDBETA 0
-3000
20

40

60

Time delay accounted for

time

Time delay neglected

5000
PA

PDALFA
0
3000
PA

PDBETA 0
-3000
20

40

60

time
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/29

Dynamic Maneuvers - Overview

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/30

Curve Fits for Steady State Sideslip Maneuvers

time

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/31

Noseboom Mounted 5 Hole Probe


Flight Validation ATTAS VFW-614

0.08

Flight Validation C-160

0.084

ATTAS
0.06

0.077

Scale
factor

K
0.04

0.070
0.02
0

0.4 0.6

1.0

2.0

Machnumber
Calibration factors independent of configuration
Dr. Ravindra Jategaonkar

0.063

20

Flap

40

deg

60

Calibration factors configuration dependent

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/32

Flush Air Data System


2

Noseboom not always possible


influences the flow
incompatible with radar

12

Flush Air Data System

11

pressure ports on the a/c nose


10
measurement of the pressure distribution
determination of the airflow parameters via suitable methods
redundancy concepts and failure detection

FADS for VECTOR (X-31 follow on project)


12 pressure ports
optimized for high angle of attack
wind tunnel measurements and preliminary flight test data

Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/33

References
Jategaonkar, R. V.,
Flight Vehicle System Identification: A Time Domain Methodology,
Volume 216, AIAA Progress in Astronautics and Aeronautics Series
Published by AIAA Reston, VA, Aug. 2006, ISBN: 1-56347-836-6
http://www.aiaa.org/content.cfm?pageid=360&id=1447
Etkin, B., Dynamics of Atmospheric Flight, John Wiley & Sons, Inc. New York, 1972.
Evans, R. J., Goodwin, G. C., Feik, R. A., Martin, C., and Lozano-Leal, R, Aircraft Flight Data Compatibility
Checking Using Maximum Likelihood and Extended Kalman Filter Estimation, Proceedings, 7th IFAC
Symposium on Identification and System Parameter Estimation, York, UK, July 1985, pp. 487-492.
Jategaonkar, R. V., Identification of the Aerodynamic Model of the DLR Research Aircraft ATTAS from
Flight Test Data, DLR-FB 90-40, July 1990.
Keskar, D. A. and Klein, V., Determination of Instrumentation Errors from Measured Data Using Maximum
Likelihood Method, AIAA 80-1602, 1980.
Klein, V. and Schiess, J. R., Compatibility Check of Measured Aircraft Responses using Kinematic Equations
and Extended Kalman Filter, NASA TN D-8514, Aug. 1977.
Maine, R. E. and Iliff, K. W., Identification of Dynamic Systems - Applications to Aircraft. Part 1:
The Output Error Approach, AGARD AG-300, Vol. 3, Pt. 1, Dec. 1986.
McRuer, D., Ashkenas, I., and Graham, D., Aircraft Dynamics and Automatic Control, Princeton University
Press, Princeton, NJ, 1973.
Mulder, J. A., Jonkers, H. L., Horsten, J. J., Breeman, J. H., and Simon, J. L., Analysis of Aircraft Performance,
Stability and Control Measurements, AGARD LS-104, Nov. 1979, Paper No. 5.
Parameswaran, V., Jategaonkar, R. V. and Press, M., Calibration of Five-Hole Probe from Dynamic and Tower
Fly-By Maneuvers, Journal of Aircraft, Vol. 42, No. 1, 2005, pp. 80-86.
Dr. Ravindra Jategaonkar

AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006

Measurements/34

Anda mungkin juga menyukai