05-A Efremov DGLR Manching 081112

7/30/2019 05-A Efremov DGLR Manching 081112

1/37

1

Prof. A.V. Efremov, Ph. D,Dean of Aeronautical school

Manching, EADS, Germany, 11-13 November, 2008

MOSCOW AVIATION INSTITUTE

MOSCOW AVIATION INSTITUTE EXPERIENCE ON

PILOT-IN-THE LOOP INVESTIGATIONS


2/37

2

Content

1. MAI fundamental investigation on pilotvehiclesystem

2. MAI applied results in manual control area


3/37

3

MAI fundamental investigationson pilotvehicle system (PVS)

database of knowledge ;

Goals: 1. Development of reliable basis

technique;

models

for investigation of single loop,multiloop, multimodality ,systems (stationary and

unstationary).2. Use of the basis for solution of applied manual

control tasks:

flight control system design;

FQ prediction

display design;

PVS parameters

Specification

DisplayControlledelementdynamics


4/37

4

Developed technique for experimental investigations:

(frequency, spectral, integral characteristics of pilot, closedloop,

openloop system )

Unified Fourier coefficient technique

Technique for preliminary definition of CooperHarper

scale metrics in groundbased investigation


5/37

5

Experience in pilotvehicle systeminvestigations exposed the following problems:

limited potentialities of wellknown mathematical pilotmodels for description of experimental data received withreal input spectrums and aircraft dynamics

considerable influence of different factors and task variables

on ground based evaluation of pilotrating and on PVScharacteristics


6/37

6

Examples

1. Pilots adaptation in low frequency range

2. Pilots ability to generate complicated actions in crossoverfrequency range

1,00,1 10 , sec-1-20

20

40

0

dB

W ,

LAHOS 1.4LAHOS 2.10

1,00,1 10 , sec-1-20

20

40

0

dB

W ,

=i

1.5 sec-1

=i 0.5 sec-1

( ) ni

ii

KS 22 +

=

1,00,1 10 , sec-1

-20

20

40

0

dB

W ,

= sec 0 = sec 0.3

sekdB

d

Wd

C

p

40lg

lg


7/37

7

3. Disagreement between groundbased and inflightsimulation

PR

ground

PRflight

averaged results

1

2

3

4

5

6

7

8

9

10

1 2 3 4 5 6 7 8 9 10

2-B

2-1

2-5

2-7

2-8

3-D

3-1

3-3

3-6

3-8

3-12

3-13

4-1

4-2

5-1

5-9

5-10

5-11

disagreement between the results in I andIII levels of pilot ratings,

decrease of pilot rating intervalPR = PRworst PRbest in ground-

based simulation,

decrease of sensitivity of flying qualitiesestimation in ground-based simulation toFQ change.

d [sm] 0.5 1.0 2.0

r [dB] 8.15 7.53 2.3

PR 8.5 8.0 3.5

e d

4. Influence of motivations (requirement to the accuracy on PR)

5. Influence of additional channel(s): singleloop taskPR = 2

in multiloop taskPR = 5

PRg

PRf


8/37

8

Problems and tasks:

determination of optimal aircraft dynamics;

exposition of regularities in pilot evaluation of Flyingqualities;

understanding of complicated behavior and ways for its

simplification; definition of rules for taking into account the different

factors.


9/37

9

Optimization of aircraft dynamics for each piloting task

Technique for definition of Wc opt Pilots limitations (PL)

),( PLtaskoptcW

task

Application of optimal aircraft dynamic

1. Development of criteria for prediction of flying qualities.

2. Agreement between groundbased and inflight investigations.

3. Flight control system design.

Solution of problems and tasks


10/37

10

Regularities in FQ evaluation1. Agreement between CooperHarper pilot rating (PR) and WeberFechner

2. Pilot workload and pilot-vehicle system parameters correlated withPR.

Data base:1. Neal Smith2. Have PIO3. LAHOS

1

3

5

7

9

1 2 3 4 5

PR

d

PR=1+5.36 ln( d )

variability PR

),( prfPR

normalized resonance peak of closed loop system

optCW

r

rr

optCCWpWpp

max

p

B

pw

e d


11/37

11

3.Relationship between CHPR and PIOR

PIOR = 0,5PR + 0,25

PR = 3,5 PIOR = 2

PR = 6,5 PIOR = 3,5

PR = 9,5 PIOR = 5,0


12/37

12

),max( PRPRPR

4.Evaluation of FQ in multichannel task

PR


13/37

13

Conclusion: Random valuePR has to be characterized by binomial law

PR

max PR = 3 5

Configuration

10

9

8

7

6

5

4

3

3D

4.1

5.1 3.3

5.11

3.13

2

1

5. Distribution of PRPilot actions variability pilot rating variability

Ex. No 1 PR = 6

Ex. No 2 PR = 9

PR random value

Peculiarities of random valuePR:

PR whole number

PR a number contained

in the limited set ofnumbers

p(PR) = C9PR1pPR1(1 p)10 PR

p =PR 1

9

PR = (PR 1) (10 PR)9

C9PR1 =

9 !

(PR 1) ! (10 PR) !

1cW

1cW


14/37

14

EXPERIMENTAL TEST ON POSSIBILITY TO USE BINOMIAL LAW FOR

DESCRIPTION OF PILOT RATINGp(PR)

ConfigurationsNumber of experiments

PR

2.122

2.86

4.1 3.8 3.8 3.12 5.10 Total22 24 20 19 17 124

2.75 3.1 3.7 6.4 7.35

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

1 2 3 4 5 6 7 8 9 10

PR

Binomial law

Experiment

PR


15/37

15

)()()(1)()()(

sWsaWsWsWsWsW

nmf

ad

p

nmvisp

ep +=

Measurements of a set of

characteristics

)();(W);( ,,ad

p, jk

e

pjkjk

vis

p jWjjW

Exposed regularities:

1. Pilot uses additional cues

2. He does it more actively when 1C11 WWwhere, WWWW Cf ==

Understanding of reasons of complicated pilot behavior

nmW

cW+

di

faW

visp

W

adp

W

1W

cW

-


16/37

16

Mathematical modeling

a. Modified structural approach

complicated form ofFVIS

the different procedure for the choice of parameters: c, FVIS, FPF

New features:

dependence of neuromuscular system on PVS task variables

taking into account pilot remnant


17/37

17

b. Modified optimal control model

recommendation for the choice of weighting coefficients

modified model of remnant spectral density

Predictor

Human operator model

Disturbance

L*

Display

Time

delay

Kalmanestimator

Vehicledynamics

V (t)u V (t)y

U (t)c

u(t)

x(t) y(t) = C (t) + D (t)x

Y (t)px(t) x(t - )

1

T s + 1N

u

New features: modified cost function


18/37

18

Composite approach to pilot modeling basedon Neural network

Stages for development of Composite model:

development of pilot neural network models (NNM)

{ } { })()( jWjW ii cp development of composite model based on pilot NNM allowed

to predict PVS characteristics


19/37

19

1. Selection of technique and definition of parameters for pilot training:

NNP /MATLAB, inverse distribution technique

Stages for development of neural networkpilot model

2. Definition of training set 2400 points3. Definition of model structure:

architecture type of model: Time Delay Neural Network (TDNN) type,

set of inputs for model: for linear Wcei(t), ci(t),yi(t),for nonlinear system e(t), c(t),y(t),

numbers of layers, numbers of neurons, types of neuron

actuation functions:

for linear controlled element dynamicsactuation function

is linear (F= 1),

for nonlinear controlled element dynamicsactuation function

is nonlinear.

G f i


20/37

20

w1

w2

w3

w4

w5

w6 - b

e t e t ( -0.40)- ( -0.45)

e t e t ( -0.45)- ( -0.50)

y t y t*( - )- *( - - 0.10)y yy

e

y*

y t y t*( - ) - *( - -0.20)

y

-0.10

y

y t y t*( - ) - *( - -0.30)y-0.20 y

e t( -0.25)

c t( )

T sy +1

1

( , ) ( )i cw b f W =

General structure of pilot neural network model

-4

-2

0

2

4

0 5 10 15 20 25 30 35 40 45 50t im e c( )

y

Comparison of mathematical modeling with experiment

15

20

10

5

0

-50

0

-100

-150-200

-25010

-1

10-1

100

100

10+1

10+1

(1/c)

(1/c)

W

experiment

NNPM optimal model structural model

experiment model

Composite approach for prediction


21/37

21

Composite approach for predictionof pilot dynamics response

1.Selection of the configuration Wck

(j) Wcm

(j) {Wci

(j)} close to Wc(j):

2.Calculation of composite pilot model WP corresponding to WC

3. Development of pilot neural network model

-50

-100

-150

-200

-250

0

50

10-1

100

10+1

(1/c)

20

40

0

-20

-4010

-110

010

+1

(1/c)

W

10075 ( )

24075 ( )

(experiment)

(experiment)

experiment

interpolation

composite model

[ ] [ ]=

+=

n

k

kckickckic AAJ

1

22)()(

180)()(

);()()(

)()()()(

);()()(

)()()()(

ikF

imFikF

imikikip

ikAimAikA

imikikip

III

III

AAAA

=

=

)()()(

)()()(

icixcixF

icixcixA

FFI

AAI

=

=

( )ie tInverseFourier

transform

NeuralNetwork

model

( )A

( ) ( )ic t( )iy t


22/37

22

MAI investigations on appliedmanual control tasks


23/37

23

1. Criteria the requirements to pilot workload and pilot-vehicle system characteristics

Potentialities: Prediction of FQ level

2. Criteria the requirements to FQ by calculation of PR

Potentialities: the possibility to define a value of PR for the selection of FQ

FIRST TYPE OF CRITERIACriteria for prediction of FQ and PIO tendency Criteria for prediction of FQ in longitudinal

in longitudinal angular motion path motion (refueling task)

Definition of and W:

Experiment Mathematical modeling (optimal or structural approach to PVS modeling

I. DEVELOPMENT OF CRITERIA FOR PREDICTIONOF FLYING QUALITIES (FQ) AND PIO TENDENCY

r

-2Pilot phase compensation


24/37

24

ControlledObject

Proprioceptive Feedback

Neuromuscular SystemCentral

ProcessingTimeDelay

Visual Block

se 0

-

+

-

C en

e

S

PFF

NMFVISF ACF

Um

Criteria as a requirements to HQSF

Level 2

Level 1

109876543210

, rad/sec

0

1

2

3

4

5

6

HQSF

15 deg/sec

150 deg/sec

Modified levels of HQSF

Level 2

Level 1

109876543210

, rad/sec

0

1

2

3

4

5

6

HQSF

Original levels of HQSF

L

m

Kj

C

UHQSF

1)( =

SECOND TYPE OF CRITERIA


25/37

25

1

3

5

7

9

11

-1.2 -1.1 -1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0Ln( e)

P R

PR=11*(1+Ln(e))

.2.1

II

I

SECOND TYPE OF CRITERIA

I group of configurations:

PR= f(e)

II group of configurations:

PR = f(pilot workload)

Criteria

PR = max (PR, PR)

Prediction of PR by mathematical modelingStructural pilot model Wp(j ) Optimal pilot model

2.3

2

2.7

3.5

1

1.5

2

2.5

3

3.5

4

4.5

1 1.5 2 2.5 3 3.5 4 4.5PR

PR

2_1

3d

4_1

5_1

)68.14.0ln1(11e

PR ++= ( )14(11.0PR += p)126.1052.0ln1(11

ePR ++= ( )0.952(11.0PR =

1

1.5

2

2.5

3

3.5

4

4.5

5

1 1.5 2 2.5 3 3.5 4 4.5 5

PR

PR

1b

1c

1d

2d

2c

2f

7c

Criteria for FQ prediction in pitch tracking taskExposed regularities

PR = max (PRa, PRb )

PRa = F (a); PRb = F (b)

DEVELOPMENT OF CRITERIA FOR PREDICTION


26/37

26

From:

J.R.Wood

AIAA-83-2105

DEVELOPMENT OF CRITERIA FOR PREDICTIONOF FQ IN LATERAL CHANNEL

Problem: Disagreement of FQ requirements developed in ground and in-flight simulation

Reason: Lateral acceleration caused by rotation

:Suggestion [ ]visPRaccPRfPR ,=[ ]

visPRaccPRPR ,max=

)(ln

)(ln

ynvest

vis

fPR

fPR

=

=

CRITERIA FOR PREDICTION OF FQ IN DUAL CHANNEL


27/37

27

CRITERIA FOR PREDICTION OF FQ IN DUALCHANNELCONTROL TASK

= PRPRPR ,max

PR, PR ratings from dualchannel system investigations

optPR

)(

)(ln36.51,

+=

)( ,opt

)( from pilot optimal control model corresponding to

dualchannel system

exp

PR

mod

PR1

2

3

4

5

6

7

8

9

10

1 2 3 4 5 6 7 8 9 10

ModelingExperiment

II. AGREEMENT BETWEEN GROUNDBASED AND INFLIGHT


28/37

28

Reasons of disagreement

In the first level of PR) a noise of estimation process due to inaccurate simulation of the different factors of flight,

b) The wrong (absence) instructions about the Cooper-Harper metrics

In the third level of PR

inability to simulate the stress situation typical for 3 level

II. AGREEMENT BETWEEN GROUND BASED AND IN FLIGHTINVESTIGATIONS ON FLYING QUALITIES ESTIMATION

1

2

3

4

5

6

7

8

9

10

1 2 3 4 5 6 7 8 9 10

2-B

2-1

2-5

2-7

2-8

3-D

3-1

3-3

3-6

3-8

3-12

3-13

4-1

4-2

5-1

5-9

5-10

5-110

1

2

3

4

5

6

7

8

9

no motion motion LAMARS MS-1 TL-39 In-f light

PR

NASA VMS

W L

Calspan

MAI

PR=PR-PRPR = PRworst PRbest

THE WAYS FOR ACHIEVEMENT OF AGREEMENT


29/37

29

THE WAYS FOR ACHIEVEMENT OF AGREEMENT

Definition of Cooper-Harper scale metrics on base of developed technique for calibration,

Simultaneous estimation of PR in longitudinal and lateral channels,

Increase of 3D objects on simulated visual scene (for the landing task)

Workstation

1

2

3

4

5

6

7

8

9

10

1 2 3 4 5 6 7 8 9 10PR

PR

..

2-1

2-5

2-7

3-D

3-6

3-8

3-13

4-1

4-2

5-1

5-9

5-10

Result : increase of PR interval from PR=2.5 up to PR=8

1

2

3

4

5

6

7

8

9

10

1 2 3 4 5 6 7 8 9 10

PR

PR. .

2_1

2_5

2_7

3d

3_6

3_8

3_13

4_1

4_2

5_1

5_9

5_10

With full set of metrics

desd Wd

add dopt

ad

opt

ad

opt

des ddd

ddFromWFL =

35

Without metrics

1

3

5

7

9

1

PR

d, sm

PR=1+5.36 ln( d )

- variability PR

5ddes dad

THE AGREEMENT BETWEEN IN FLIGHT AND GROUND BASED


30/37

30

INVESTIGATIONS FULFILLED ACCORDING THE DEVELOPED TECHNIQUE

2.5 m/s 78 m 150 mAdequate

1.5 m/s 1.5 m 75 mDesired

Touchdownvelocity

VTD

LateralerrorY

LongitudinalerrorX

0.5 1.8m/s

Less then 60% radius of basketAdequate

0.9 1.4m/s

Less then 40% radius of basketDesired

Contactvelocity

LateralerrorY

LongitudinalerrorX

1.5 mradAdequate

5.0 mradDesired

Angular error

Landing

Refueling

Aimtoaim tracking

III. Means for improvement of pilot actions and FCS system conjunction


31/37

31

p p y j

Goal: To suppress exposition of flight control system limited potentialities

WAYS FOR SOLUTION OF PROBLEM:

MANIPULATOR WITH VARIABLE STIFFNESS

LOGIC OF SYNCHRONYZED PREFILTER TO SYNCHRONIZE PILOT ACTION AND FLIGHT CONTROL

WITH LIMITED POTENTIALITIES BY LINEARIZATION OF PILOTAIRCRAFT SYSTEM CHARACTERISTICS

SYNCHRONIZED PREFILTER

Kf 1/s

1.

2. Kfo

max

max&

&

law 1:

restoration of initial gain coefficient Kfo

law 2:quick changeof Kf

law 1:

restoration of initial gain coefficient Kfo

law 2:quick changeof Kf

Kf 1/s

1.

2. Kfo

max&max&

&

1)1(

++=

TapTP

XP

X

W1nonlinear standard prefilter

Additional forceregulation law

1 1/pKf


32/37

Effect of manipulator with proposed regulation of force


33/37

33

Effect of manipulator with proposed regulation of force

constPX =

variable force

After failure

Control surface deflection

(limiter output) Px=const

(limiter output) variable stiffness

Px=const

variable stiffness


34/37

34

IV. Some aspects of direct lift control use

Direct lift control surfaces flapperrons, canards

Direct lift control allows to:

improve short period dynamics

conserve flying qualities in caseof FCS failure

to suppress the speed instability

INVESTIGATION OF DLC EFFECTIVENESS IN REFUELING


35/37

35

Improvement of accuracy

Aircraft with DLC

Without DLC

Effectiveness of direct lift control in carrier landingDLC allowes to suppres speed instability in carrier landing


36/37

36

DLC allowes to suppres speed instability in carrier landing(because =const)

Without DLCWith DLC

Results of experiments

0

2

4

6

8

10

Without DLC

With DLC

Touchdown

0

10

20

30

40

50

3.5

47

22 , X

PR

V Additional information on path angle


37/37

37

V. Additional information on path angle

Integration of DLC and path angle indication gives an improvement ofperformances up to 20 30%

With Without

Documents

05-A Efremov DGLR Manching 081112