challenges in flight control design development and ... techniques/ppt/v patel... · challenges in...

Challenges in Flight Control Design Development and TestingDevelopment and Testing

Dr. Vijay V. Patelj yNational Control Law Team

Aeronautical Development Agency, Bangalore

Flight Control Theory and PracticeS i i lik d l h d hi h h l• Scientists like to develop methods which have general applicability

• Flight control system has evolved over a century and it• Flight control system has evolved over a century and it is safety critical for modern high performance aircrafts.

• Flight control engineers are extremely conservative g g y(would like to use proven design techniques)

Respect the Unstable

The practical, physical (and sometimes dangerous)consequences of control must be respected, and theunderlying principles must be clearly and well taught

Overview

• The Airframe• The Control Law Design Cycle• Ground Testing and Validation• Flight Testsg• Design Challenges• The Way AheadThe Way Ahead

Salient Features of LCALight weight, single engine, multirole supersonic aircraft with all weather day / night capability

Tail‐less compound delta airframe configuration uses extensive carbon composites and co‐cured co‐bonded technology for realizing low weight and high strength

Relaxed longitudinal stability with scheduled leading edge slats for high alpha maneuvering and combat

Quadruplex all digital fly‐by‐wire flight control system with built‐in redundancy management system and re‐configurable control laws ensures safety of flight with agility and maneuverability

Oct 06, 2013 Pravartana 2013 TEQIP Workshop

The SensorsThe Sensors

Vertical Datum :X=0

+ Y

+ X

A/C C/L ; Y=0

1 Nose Airdata Probe

FRL : Z=0

1 2 3 4 5 62 Side Airdata Probes

3 AoA VanesFRL : Z 0

+ Z

4 Acceleration Sensor

5 Temperature Sensor

+ X

6 Rate Sensor

The Air Data SystemThe Air Data System

Total Air Temperature

sensor

Side air data probe

AoA VaneNose air data probe

Why Relaxed Static Stability ?LTLT

LTα

WWD

L

α

LT

FTFT

UnstableStable

G i i if

LIF

LIF Gain in Lift

Decrease in Drag

DRAGAngle of attack (α)Oct 06, 2013 Pravartana 2013 TEQIP Workshop

Limits on Level of InstabilityLimits on Level of Instability

• Hardware (actuator, sensors etc.) capabilities poses restriction on level of

Kq(Tθ2s+1)(s/σ - 1)(s/ τ + 1)

ω2ζ ω s + ω2s2 + 2-

+

- Hardware Plant

Outputinput

pinstability– if K1 > K2, aircraft can

never be stabilised

KGain

Hrd Modenever be stabilised– more stringent

requirement for t i i

X

XX O

K2CH

Hrd. Mode

guaranteeing margins i.e. for ±6 dB gain margin 1/2 K1 ≤ 2Κ2 X

XX OK11/Tθ2 σ

CHHrd. Mode


Key Elements which affect the yperformance of FCS

Airframe

Control Laws and associated HardwareControl Laws and associated Hardware

Pilot Seat location

Stick Characteristics and Type

DisplaysDisplays

Operating Environment (Cockpit / External)


Vi l C

Human Pilot Modeling & Pilot – Aircraft Coupling

Pil t i fyu

Visual Cues

Pilot Aircraft

Attitude

y

C nt l FAcceleration

Control Force

Motion CuesPilot Model

K= 1-100; D=0.2 secTA= 0-2.5 sec (lead)

Pilot Model

)1()( DsesATKsy −+ A ( )TL= 0-20sec (Error smoothing Lag)TN≅ 0.1 sec ( Neuro muscular lag))1)(1(

)1()()(

sNTsLTesATK

susy

+++

=

Control Law RequirementsControl Law Requirements

• To Recover Stability and Provide GoodTo Recover Stability and Provide Good Handling Qualities to the Pilot– Stability Robustness y

• Guarantee the required stability margins– PIO Resistance

• Ensure good stability margins with pilot in the loop– Performance Robustness

I i t ith t t d i f l• Invariant response with respect to aerodynamics, fuel, etc.


Margin RequirementsRequirements (Based on MIL‐SPEC‐9490D)

10 Plots of Closeness ParameterGain

Rigid4

6

8

10 Plots of Closeness Parameter

Phase

-2

0

2MarginsRequired

Gai

n (d

B)

10

-8

-6

-4

Nichols PlotStructural -180 -170 -160 -150 -140 -130 -120

-10

Phase (deg)

•For computational tractability the trapezium is approximated by an ellipse This


•For computational tractability, the trapezium is approximated by an ellipse. This leads to the definition of a metric, Closeness Parameter (CP) that incorporates both GM and PM.

Control Law Requirements Contd…Control Law Requirements Contd…

– Level 1 Handling Qualities• Crisp and Predictable Response to the PilotC f H dli– Carefree Handling

• In respect of AoA, Sideslip, Nz, Inertia Coupling, etc.– Sufficient Structural Stability MarginsSufficient Structural Stability Margins

• No coupling with structural modes– Autopilot Modesp

• For ease of delivering stores, prevent disorientation, etc.


Quadruplex Digital FCS - Architecture

MILMIL--STDSTD--15531553 RH LE Slat *

* - Actuators

MILMIL STDSTD 15531553 RH LE Slat

RH O/B Elevon *

RH I/B Elevon *

Rudder *

Rate Gyro(p, q, r)

DFCCDFCCAccelerometer (Ny,

Nz)LH I/B Elevon *

LH O/BElevon *

Air Brake *

AoA Sensor

LH LE Slat *

Crash DataFCS Analog / Discrete Signal

S i l Di i l I f

Air Data (Ps, Pt)

Serial Digital Interface

FCS Analog /

Disc FTI Signal

FTI

GUH Panel

HQPACKMathematical Model of

Aircraft Mil Specs

TEJAS IFCS TEST FACILITIES

Correlation withPilot Rating

Graphical Output

Mil Specs

Analytical HQR & PIOR

Control Law Dvpt Environment (NAL)

Engineer-in-loop Simulator (NAL) Pilot-in-loop Real Time Simulator(ADE)

Structural Coupling Test Facility (HAL)

Lightning Test Facility (CABS) Non Real Time Test Facility (ADA)g g y ( ) Non-Real Time Test Facility (ADA)

Aircraft Tests (HAL) IFCS Facility(ADA) Iron Bird(HAL) Mini Bird(ADE)

Multi-disciplinary modeling & Simulation Requirementsfor Fly-by-Wire Control Law Development

Aerodynamics Control

for Fly by Wire Control Law Development

AerodynamicsFlight

Dynamics

ControlEngineering

FlightC nt l S st m

Human PilotDynamics

Control System

StructuralDynamics

Landing

DynamicsPropulsion

Landing GearHydraulics


Design Requirements

StabilityHQThe Design Cycle

REFINE

Linear Models

Linear Control

Control Law Design & HQ Analysis tools

FCS Constraints etc

N

REFINENon‐linearControl

AERO DATA BASE

D

AT

A

Control Designs

Design O K ?

NO

NO6 DOF

Control Shaping

YES

EN

GIN

E

MASS / INERTIA

Design O.K.?

O.K.?6 DOF Simulation

YES

e (k

m) Non Real

TimeReal Time (‘ELS’)

Real Time (‘RTS’)

YES

ValidatedC

Alti

tude Time ( ELS ) ( RTS )

In-flightSimulation

Control Law


Flight Control System Structure

Visual and Motion CuesControl Decision Visual and Motion Cues

+

+Forward CLAW

±6dB-350

Mil Spec F9490DRigid

Flexible

Handling QualityMil Spec 1797

Aircraft Rate & accln

sensors

Actuators

- 8dB (small gain stabilization)

Pilot Input

+

Feedback CLAW

Notch Filters

)Mil Spec F9490D

Offline Design (Table lookup fn of Mach and altitude)

Air-data

Gain SchedulingVaneCarefree

Maneuvering

Oct 06, 2013 Pravartana 2013 TEQIP Workshop Airdata Sensors

Air-data computation +

g+ Low Speed Flying

CLAW

CLAW Reversionary ModesCLAW Reversionary Modes• AoA failure

– An equivalent AoA signal is generated for feedback by processing total normal acceleration in addition to suitably reducing the command authority.

• Airdata system failure – A fixed gain reversionary mode as an ultimate backup. – Gains are frozen at the last good value, – Pilot can then manually switch over to the standby gains using a switch in the

cockpit. – Two sets of fixed gains in the standby mode, which get automatically selected

depending on the status of the undercarriage lever.

• Elevon Failure


Longitudinal Feedback Controller Schematic

Gain &Low-Pass Filter

PFC10

Scheduled Gain

Nz Anti-aliasing & Notch Filters

1g cosθ

≅ ΔαNz-alpha conversion gain+

- +

+

Gain andWashout Filter

PFC10

Scheduled G i

Low-Pass FilterPST05q

Anti aliasing &

≅ Δα

Filters

+

+

+

+

Scheduled Gain and Phase Advance Filter

Gain

Scheduled Bias

Anti-aliasing & Notch Filters

≅ α.

Washout FilterPST05

+

+ ToScheduled Gain and Phase Advance Filter

Gain

+

+

++

+

+

To

de

α Signal Conditioning Filter

Bias

Low Pass Filter

Scheduled Bias

Nz-alpha conversion gain

≅ α

+α

+

+

Pt-Ps Signal Conditioning Filter &

Low Pass Filter

+

-

Longitudinal Axis Command PathLongitudinal Axis Command Path

From Feedback

Small Amplitude

Pre-Filter

Trim Correction

Small Amplitude

Pre-Filter

Trim Correction

Pitch StickShaping & Limiting

Amplitude

LargeAmplitude

Pre-Filter

Rate-LimiterDesensitizer

δeSaturation and Rate-Limiter

Pitch StickShaping & Limiting

Amplitude

LargeAmplitude

Pre-Filter

Rate-LimiterDesensitizer

δeSaturation and Rate-Limiter

Auto-Pilot

Limits as a function of Pitch Stick Command

No-Windup Integrator

Pitch Trim

Auto-Pilot

Limits as a function of Pitch Stick Command


Pitch Trim

Lateral/Directional Controller Schematic


Roll Stick

Roll Trim LimiterPitch Stick Position

Lateral

δa

p Anti-aliasing and Notch Filter

Low-Pass Filter Phase Advance

Saturation and Rate-Limiter

Roll Stick Shaping & Limiting Rate-LimiterDesensitizer Pre-Filters Gain As a function of

Pitch Stick Position

β

Signal Conditioning Filterr

≅β.

Anti-aliasing and Notch Filter

Estimate βand

β.

Estimated α

≅β

Gain and Phase Advance

GainDirectional

Ny Low Pass Filter

Filter

Anti-aliasing and Notch Filter

Pitch Stick PositionRoll Stick to Rudder Interconnect Gain

Aileron to Rudder Interconnect Gain

Rudder Pedal Shaping & Limiting

Rate-Limiter Saturation and Rate-Limiter

δr


Yaw Trim Limiter

Gain As a function of Pitch Stick PositionPre-Filter

g

Air Data Signal Consolidation and Leading Edge Slat ControllerAir Data Signal Consolidation and Leading Edge Slat Controller

P

Air Data Signal Consolidation

Ps Low Pass Filter and Rate Limiter

Pt-Ps Low Pass Filter and Rate Limiter

÷ √

SP

For Air Data Gain Scheduling

√

÷ √MP

Slat Authority Limiters as function of

Leading Edge Slat Control Law

Normalised slat Command

Pitch Stick Shaping & Limiting

Saturation andRate-Limiter

Washout Filter

Dynamic Pressure and Mach

Hysteresis

Low – Pass Filter

Processed α

Kinematic coupling of AoA and AoSS


Control Law ValidationControl Law Validation

CostIn flight

PrototypeAircraft LCA

Iron Bird

In-flightSimulator

HAL

CALSPAN

Iron Bird

Engg.Simulator ADE, BAe

HAL

“Design”Simulator

Simulator

NAL

Realism

Fixed Base Real Time Flight SimulatorELS ‐ NAL

Important Requirements of RTSImportant Requirements of RTS• Accurate DynamicsAccurate Dynamics

• Realistic Delays• Realistic Delays

• Adequate Cueing for Task

• Feel System Dynamics

Importance of Pilot in theImportance of Pilot in the Loop Simulation

Visual and Motion Cues

AircraftCLAWPilot

Sensor Feedback

Man-Machine Interaction

In‐flight SimulationIn flight Simulation

Ironbird SchematicIronbird Schematic

DFCC

Test Rig Consists of Real Actuators, Undercarriage,

H d li SReal sensors

Hydraulic System

I/O RACK

Test Rig

Cockpit DH-EUADA-FDS

I/O RACKAircraft Simulation

Cockpit DH EUADA FDS Computer

Typical Test for RM LogicTypical Test for RM Logic

>Tt1/2

t /2

Ch.1t1/2Ch.2

Ch 3-t1/2

Ch.3

Ch.4-t1/2

-t /2RM Selected -t1/2Selected Value

Ai ft Ri id B d & St t l M d D iAircraft Rigid Body & Structural Mode Dynamics

20

0

itude

(dB

)

Total

Design BW

-40

-20

Mag

ni

RigElastic

Rigid Body Modes

0.1 Frequency in Hz

1.0 10.0 50.0idic

Rigid Body Modes

0.1 Frequency in Hz

1.0 10.0 50.0

Rigid Body Modes

ASE ModesPilot DynamicsActuators

Rigid Body Modes

ASE ModesPilot DynamicsActuators


ActuatorsActuatorsHigh AOA Instabilities

Elevator Actuator MarginElevator Actuator Margin

A/Cde

STIMMeasureqnzA/C

da

dr

nzprndr ny

DFCC


Gain Margins –l f d hSimplified Schematic

DFCC H/Wde

A/C

q

DFCC H/WA

A/CNz

q Notch Filtersq CLAW

Nz Notch FiltersNz CLAWB

A B


,1deAL =

deBL =2

A/de

I/PO/P qnz

Notch Filter Design based on SCT Experiments

A t ti h A/C

dadr

nzpr

ny

A systematic approach was evolved to design notch filters for the various sensor

Rigid Body Control Law

Notch Filters

paths in the presence of coupling between structural responses of all three axes of

q CLAW q NF

DFCCresponses of all three axes of the aircraft.

A difficult optimisation

P11

P12 P21qde

pproblem was converted into five independent simple optimisation problems P12 P21

P22 pda

optimisation problems without introducing conservativeness.

p CLAW p NF“A Systematic Procedure to design notch filters for structural Responses with multi-axis coupling”, AIAA Journal of Guidance, Control and Dynamics, Vol. 22, No.2, pp. 349-357, March-April 1999.Oct 06, 2013 Pravartana 2013 TEQIP Workshop

1 0B o d e D ia g r a m

-1 0

0

gnitu

de (d

B)

- 3 0

-2 0Mag

9 0

1 3 5

-1 3 5

-9 0

-4 5

0

4 5

Phas

e (d

eg)

Ac tu a l

c o m p u te d

F r e q u e n c y (H z )1 0 -1 1 0 0 1 0 1 1 0 2 1 0 3

-1 8 0

1 3 5 c o p u te d

Bode’s 1 udM∞ ⎛ ⎞Gain-PhaseRelation

c 21( ) ln coth

udMw dudu

φπ −∞

⎛ ⎞= ⎜ ⎟

⎝ ⎠∫

Where, u=ln (w/wc) Oct 06, 2013 Pravartana 2013 TEQIP Workshop

Control Law Design & ValidationControl Law Design & Validationgg

Prototype

CostIn-flightSimulator

ypAircraft

Iron Bird

Simulator

Engg.Simulator

“Design”Simulator

Oct 06, 2013 Pravartana 2013 TEQIP WorkshopRealism

Objectives of Flight TestsObjectives of Flight TestsValidate the performance of the FCS in flightValidate the performance of the FCS in flightEstablish the handling qualities of the aircraftCalibrate the airdata systemCalibrate the airdata systemIdentify the aerodynamic stability and control characteristics and establish the correlation of flight responses with the Real time simulator responsesVerify Stability Margins in Flight


Online Gain and Phase MarginsOnline Gain and Phase MarginsP2 qδe

A

Pitch Stick Path Filters

Pstick

BPlant

NzActuator

ControllerP1

20

de to de -Transfer Function for F8_seg4

From flightFrom Linear ModelG:= TF from P to P

5

10

15

ude

in d

B

G:= TF from P1 to P2 L:= TF from A to B

LG

−=

11

-10

-5

0Mag

nitu

GGL 1−

=


-200 -180 -160 -140 -120 -100 -80 -60 -40 -20 0-15

Phase (deg)

Modern Robust Control Applied to LCA lControl Laws

• Major Bottleneck:– Classical Gain and Phase

1

20

Margins still the onlyRobustness Criterion used

Open loop transfer function

0

5

10

15

itude

indB

Margin Template

for Flight Controllers (MIL Spec)

-15

-10

-5

0

Mag

ni

– Led to conservative H∞/μcontrollers

.

-250 -200 -150 -100 -50 0-15

Phase (deg)


Robust, fragile, or optimal?Robust, fragile, or optimal?, Keel, L.H.; Bhattacharyya, S.P.IEEE Transactions on Automatic Control, Volume 42, Issue 8, Aug 1997 Page(s):1098 - 1105

CC+

-c yue P

Summary: We show by examples that optimum and robust controllers, designed by using theH2, H∞, l1, and μ formulations, can produce extremely fragile controllers, in the sense thatvanishingly small perturbations of the coefficients of the designed controller destabilize theclosed-loop control system. The examples show that this fragility usually manifests itself asextremely poor gain and phase margins of the closed-loop system. The calculations given hereshould raise a cautionary note and draw attention to the larger issue of controller sensitivitywhich may be important in other nonoptimal design techniques as well

Margin Requirementsg q

Gain ImRigid

Structural

Phase-1

Re

Nichols Plot Nyquist Plot

RigidStructural

yq

Margins Template Converted to U t i t th Pl tUncertainty on the Plant (Non Conservative)

|L(jω)| in dB|L(jω)| in dB

+6dB

-1800, 0 dB

∠|L(jω)|∠|L(jω)|

-1450-1800, 0 dB

∠|L(jω)|-6dB

6dB 350 template in theThe template converted to an

ncertaint on the plant6dB-350 template in the Nichols Plot

uncertainty on the plant

Margins and Complementary Sensitivity

Th B i D i P bl b th t fThe Basic Design Problem becomes that ofminimizing sensitivity for the nominal plant subjectto a bound on the biased complementarysensitivity.

10CP +c yu

z1

PΔ

0.25e1

75.01

25.11 0

0 <+ ∞CP

CP

C

-yu

P0

“A Modern Look at Gain and Phase Margins: An H∞/μ Approach”,AIAA Guidance Navigation and Control Conference, ControlTheory Analysis and Design, Boston, August 10-12, 1998, G. S.Theory Analysis and Design, Boston, August 10 12, 1998, G. S.Deodhare and V. V. Patel

Modified Mixed Sensitivity Minimisation P blProblem

– The Basic Design Problem becomes that of minimisingsensitivity for the nominal plant subject to a bound onthe complementary sensitivity of a scaled plantthe complementary sensitivity of a scaled plant.

W 1min 1)25.1( PCW∞+ CP

WC 1

min 1gstabilisin 1

)25.1(1)(

2 <+ ∞PC

Wsubject to


SolutionSolution

Th P bl “ l d” i it ti• The Problem was “solved” using an iterative procedure

+ )251(11min 1 PC

W i1

)251(1)25.1(

2 <+ PC

PCWs.t∞+ )25.1(1 PC

PCWW iii 1111

)25.1(1 −− +=

)25.1(1+ ∞PC

0 WW =where PCWW

i 111 1 −+

11

11 WW =where


ε<+

−+ ∞−∞ PC

WPC

Wii 1

11 11

11

till

Utility in DesignUtility in Design

• Helps to benchmark a classical controller and findHelps to benchmark a classical controller and find “limits of performance”.

• Identify new structures for a classical controller ydesign.

• For the multivariable case the uncertainty can be yadded to each input of the plant ‐ μ synthesis for non‐conservative designs.


Clearance of Flight Control LawsClearance of Flight Control Laws

• Clearance of control laws for flight – a very laborious g yprocess.

• Certification Authorities have to be shown that f f fl ll f h l haircraft is safe to fly in all parts of the envelope with

the nominal controller and with failures.• Safety to be ensured with all aerodynamic tolerances• Safety to be ensured with all aerodynamic tolerances and tolerances on airdata parameters.

• For all configurations of the aircraft – with different gcombinations of weapons, drop tanks, etc.


Utility of Modern Control Techniques f lfor Clearance

• Current approach is gridding of the envelopeCurrent approach is gridding of the envelope• Disadvantages‐

– Number of cases goes up exponentially as aNumber of cases goes up exponentially as a function of the parameters

– No guarantee of satisfaction of requirements at non‐grid points

– Only extreme values of tolerances considered


Utility in Flight Clearance …Utility in Flight Clearance …

• Recent research b the GARTEUR• Recent research by the GARTEUR Group has demonstrated how the

l t iti it it i fcomplementary sensitivity criterion for stability margins can be used for

l t d fli ht laccelerated flight clearance• Achieved by modeling all the uncertainties in database as LFTs and using μ‐norms to clear the flight envelope.


Air Data SensorsTotal Air Temperature

sensor

Side air d t bdata probe

AoA VaneNose air data

probe

SSA Vane

Sept 07, 2012 Presentation at IIT Kanpur

Concluding Remarks

Design of Control Systems with a human in the control loop is a very challenging problemloop is a very challenging problem

Flight Simulation plays a very important role in the design ti i tioptimization

A Good R&D Base has been established in the country to undertake

aircraft flight control design, , airdataairdata system algorithm developmentsystem algorithm developmentairdataairdata system algorithm development, system algorithm development, AerodataAerodata update and update and airdataairdata system calibration, system calibration, Flight testing and flight envelope expansion. Flight testing and flight envelope expansion.

Sept 07, 2012 Presentation at IIT Kanpur

Control Law Flying on LCA

Thank You Theory - without Practice - is Sterile; Practice - without Theory - is BlindPractice without Theory is Blind-- Karl Marx

AcknowledgementAcknowledgement

This presentation is a summary of the outstanding contributions

made by the National Control Law Team scientists from

NAL, ADA and HALOct 06, 2013 Pravartana 2013 TEQIP Workshop

challenges in flight control design development and ... techniques/ppt/v patel... · challenges in...

Documents