Title: Retrofit Reconfigurable Control of an F/A-18C
1Retrofit Reconfigurable Controlof an F/A-18C
Society of Flight Test Engineers Patuxent River,
MD 16 NOV 2005
- Tony Page
-
- Dean Meloney
- Naval Air Systems Command
2Background
- The Navy has been investing in reconfigurable
control technology as part of the Flight Control
Predictive Diagnostic project (ONR funded DI
program) - NAVAIR leveraged several SBIRs to expand research
- One SBIR company developed a novel in-line
retrofit reconfig. approach - Original plan was to flight test the method under
the Phase II SBIR with support from NASA - When NASA support evaporated, decision was made
to pursue flight test using DI funds (Technology
Push) - Limited flight test demonstration (4 flights)
3Flight Control Reconfiguration
Redistribute control commands to compensate for
battle damage or actuator failure
During a pitch maneuver for example, the ailerons
and rudders can be deflected to counteract the
roll and yaw coupling induced by the
damaged/failed stabilator.
4Motivation
Background numerous documented cases of loss of
commercial and military aircraft and life that
are attributed to major flight control system
failures
Problem the full capabilities of the aircraft
are not realized on production commercial or
military vehicles (even those with digital flight
control systems), and the pilots are often unable
to effectively fly the damaged aircraft
Motivation many aircraft have intrinsic ability
to maintain controlled flight and land despite
flight control failures, midair collision damage,
or upset conditions
Automatic reconfiguration to maintain
controllability and recover, as closely as
possible, the baseline handling qualities of the
airplane
Fault Tolerant Control Systems software and/or
hardware to enable fail-safe or fail adaptive
operation
5Status of Existing Research
- Status of the Field variety of techniques and
several demonstrations have shown promise of
software reconfiguration to handle a large number
of otherwise catastrophic upset and damage
conditions - multiple significant flight demonstrations in the
- past 10-15 years, with several involving Boeing
- (Self-Repairing FCS, PCA, RESTORE)
- VISTA F-16 flight tests (Self-Designing
Controller, 1996)
SDC Milestone first time aircraft landed under
reconfigurable control
- However significant gap between the research and
- use of the technology
- many methods cannot be applied
- to current generation or legacy aircraft
- hardware redundancy is common
- approach to fault tolerant design
- VV steps have received less attention
- Federal certification authorities
- lack the resources to evaluate and
- certify novel technology
Recent and Ongoing Research Address the issues
with designs that can be retrofitted into
existing aircraft
6Retrofit Reconfiguration Architectures
ProductionControl Law
u
?
?
Pilot Input
Sensor Data
7Comparison of Retrofit Architectures
- Both Architectures Designed to be Non-Interfering
- Nonzero inputs only result when performance
differs from baseline - Advantages of Parallel Implementation
- Control of individual actuators provides more
opportunities to reconfigure the aircraft - Advantages of In-Line Implementation
- Command limiting, structural filters, etc. remain
in effect - Safety features of existing CLAW need not be
duplicated or abandoned - Architecture is similar to autopilot
If it isnt broken, dont fix it!
More Powerful
Should be easier to certify. Lessened VV
Requirements(if being added to an existing
system)
8How Does it Work
Undamaged Airplane
- Retrofit control law has model of how aircraft
should respond - Available sensor data is used to identify in
real-time a model of how aircraft is responding - Retrofit control law compares the two models and
computes an additive command for pitch stick and
roll stick in software
Pilot Inputs
Damaged Airplane
Pilot Inputs
Damaged Airplane
Pilot Retrofit Inputs
Reconfiguration covers up damages to allow pilot
to fly airplane with minimal manual corrections
to instinctive commands
9Retrofit Control Law Diagram
Design Concept Adaptively computed gains are
applied to feedforward, feedback, and integral
error states to yield control variables (which
are increments to pilot commands that recover, to
the extent possible, nominal flying qualities)
2
3
Receding Horizon Control
System Model
Model Inputs (states, controls, airdata, etc.)
1
Ref. Model (Commanded Response)
Ffq
Commands
Fcmd
Control
S
Fi
1/s
AW
S
Fpl
States
f
1
10Major Components
- Reference Model (Prescribed Offline)
- Low order equivalent system transfer functions
from pilot - stick to aircraft responses (i.e. pitch stick to
pitch rate, - roll stick to roll rate, etc.)
- Parameter ID System Model (Adapted Online)
- State Space model with time varying parameters
- Modified Sequential Least Squares algorithm a
regularizedparameter estimation method to
determine system model terms - Model-Based Adaptive Control (Solved Online)
- Online control design procedure that operates
onthe system and reference models to generate
controlcommands that cause aircraft dynamics to
trackreference models. - Continuous-time formulation of receding
horizoncontrol for a state-space system model
1
How you want theA/C to respond
2
How the A/C isactually responding
3
How to make theA/C respond morelike what you
want
11Retrofit Control Law Testing
Batch simulation Tens of thousands of NRTCASTLE
simulation cases Pilot-in-the-loop Successful
software only pilot-in-the-loop
simulationtesting with Boeing andNavy
Pilots Hardware-in-the-loop Extensive
pilot-in-the-loop verificationof retrofit
control running real-timein the FSFCC Flight
Testing Two flights completed
Mach Number
NRT Test Point
Piloted Soft. Only Sim. Test Point
HILS Flight Test Points
12Batch Simulation Results
- Completed extensive simulation testing using the
Navys high-fidelity simulation environment
(CASTLE) - Wide range of failures and damage
- Turbulence
- Sensor Noise
- Different Aircraft Configurations
- 75 of cases rated as good or excellent with
regards to ability to restore nominal flying
qualities - 85 of cases rated as fair or better
- Majority of remaining cases did not have
sufficient control power to achieve fair or
better due to physical limitations
13Piloted Simulation Results
- Piloted Simulation Scope
- Navy and Boeing pilots
- Three flight conditions1 (0.7M, 20kft) 2
(0.9M, 30kft) 3 (0.6M, 30 kft) - Failures to primary aerodynamic control
surfaces(stab., aileron, rudder)
Bars comprise 12 to 15 HQR assessments of
refueling, target tracking, bank / heading /
pitch capture, etc.
Cooper-Harper Handling Qualities Ratings
Excellent
Fair Some Mildly Unpleasant Deficiencies
Moderately Objectionable Deficiencies
Loss of Control During Some Operations
Major Deficiencies
14Pilot Comments and Observations
- Pilot Tracking Task (Mach 0.60, 30 kft), Left
Stabilator 6 deg. Down - Close agreement between commanded achieved
pitch and roll - Data confirms pilots observation that
- Improvement was eliminating the strong right
roll-off and the roll coupling with pitch. A yaw
left with pitch up, yaw right with pitch down was
introduced. - Inflight Refueling Task (Mach 0.70, 20 kft),
Left Aileron 20 deg. Down - Close agreement between commanded achieved
pitch and roll - Uncommanded yaw significantly less for this
flight condition and task - Data supports pilot assessment of system for this
case - The elimination of the constant left stick
input and the roll coupling were a sure
improvement. It was hard to see a degradation in
tanking resulting from any yaw coupling that may
have been present. Difficult, but roughly
equivalent to the baseline airplane.
15Retrofit Algorithm Selection
-
- meaningful demonstration possible
- without pedal-augmented retrofit
- architecture
- stick-only architecture represents
- appropriate tradeoff of performance
- and hardware implementation feasibility
- in the 1750A
Substantial reconfiguration benefits shown in
piloted simulations with rudder pedal omitted
from retrofit algorithm
Reconfiguration improvements of the stick and
pedal retrofit control law are lessened
slightly because of slower update rates in the
1750A hardware
Conclusion use stick-only retrofit
architecturefor HILS and flight testing
16Flight Hardware
- Fleet Support Flight Control Computer
(FSFCC)(formerly the Production Support FCC
(PSFCC)) - Standard F/A-18A-D FCC but with an additional
processor card in each channel - During flight, control of the aircraft can be
passed from the baseline (701E) processors to the
research (1750A) processors in order to perform
an experiment - Control is passed back to the standard flight
control system in the event that any of the
multiple safety monitors are tripped (or manually
via paddle switch)
FSFCC
17Implementation of Retrofit Controller
F/A-18 Fleet Support Flight Control Computer
Military Spec 1553 Inputs
Baseline F/A-18 Central Processing Unit (701E)
Built-In Test, Executive, and Data Management
Control Laws (V10.1)
Input Signal Mgmt
Output Signal Select Fader Logic
Actuator Signal Mgmt
Surface Actuator Analog Interface
Analog Inputs
18701E Safety Monitoring
Automatic Disengage Criteria
191750A Safety Monitoring
- Checks health and status of miscellaneous
parameters - For example Spin, Spin switch, heading hold
- Envelope limits
- Monitors p, q, r, Nz, Altitude, Airspeed, etc.
- Parameters must be within predefined limits in
order to engage the research processor - Research processor will automatically disengage
if necessary - Limits are contained in a lookup table
- Pilot selects table entry through DDI inputs
20HILS and Flight Test Plan
- Conduct flight test maneuvers and evaluate
handling qualities for the following scenarios - Retrofit control inactive, no failures(provides
nominal performance baseline) - Retrofit control active, no failures(demonstrates
non-interference) - Retrofit control inactive, with
failures(provides degraded performance baseline) - Retrofit control active, with failures(demonstrat
es benefit of reconfiguration) - Failures under consideration
- Right aileron stuck at given position ( 30
offset) - Right stabilator stuck at given position ( 6
offset)
21HILS and Flight Test Plan (contd)
- Flight test maneuvers
- Stick doublets
- Pitch and bank angle captures
- Guns tracking (with chase as target)
- Aircraft configuration
- F/A-18C
- Clean with exceptionof center-line tank
- CR and PA
22Pilot TOD (Maj. Matt Doyle)
HILS Results 30 Aileron Failure (UA)
Average ?HQR 1.5
Guns Tracking
?HQR 2.0
Bank-to-Bank Rolls
?HQR 1.0
Pitch Attitude Capture
?HQR 1.5
Retrofit Smooth Mnvr. / Coarse Tracking
V10.1 CAS Smooth Mnvr. / Coarse Tracking
Legend
Retrofit Aggressive Mnvr. / Fine Tracking
V10.1 CAS Aggressive Mnvr. / Fine Tracking
23Flight Test 30 Aileron Failure (v10.1)
24Flight Test 30 Aileron Failure (Retrofit)
25Pilot TOD (Maj. Matt Doyle)
Flight Results 30 Aileron Failure (UA)
Average ?HQR 1.6
Guns Tracking(Maneuvering)
?HQR 1.0
Guns Tracking(Level Turn)
?HQR 2.0
Bank-to-Bank Rolls
?HQR 1.0
Pitch Attitude Capture
?HQR 2.5
Retrofit Smooth Mnvr. / Coarse Tracking
V10.1 CAS Smooth Mnvr. / Coarse Tracking
Legend
Retrofit Aggressive Mnvr. / Fine Tracking
V10.1 CAS Aggressive Mnvr. / Fine Tracking
26RD Summary
- Tens of thousands of NRT CASTLE
- simulation cases
- Successful piloted simulations with
- Boeing Navy pilots
- Implementation in US Navy F/A-18 FSFCC
- Successful HILS piloted testing
- Half way through flight test program
F/A-18 flight testing at Patuxent River NAS
27Back-up Slides
28Open Literature Publications
Barron Associates, Inc. Ward, D. and Monaco, J.,
"System Identification for Retrofit
Reconfigurable Control of an F/A-18," AIAA
Journal of Aircraft (to be published). Monaco,
J., Ward, D., and Bateman, A., "A Retrofit
Architecture for Model-Based Adaptive Flight
Control," AIAA Paper No. 2004-6281, in Proc.of
AIAA Intelligent Systems Conference, Sep.
2004. Boeing and NAVAIR Black, S., et. al.,
"Reconfigurable Control and Fault Identification
System," 2004 IEEE Aerospace Conference. March
6-13, 2004, Big Sky, MT.
29NAVAIR Program Background
- The Navy has been investing in reconfigurable
control technology as part of the Flight Control
Predictive Diagnostic (FCPD) project - FCPD Objective To develop demonstrate damage
failure diagnostics/prognostics approaches for
reconfigurable control, condition-based
maintenance, and improved situational awareness
Health, Faults, Anomaly
Aircraft Level
Component Health
Faults, Health Observable at A/C System Level
Damage ID
Component Status
Ability to perform in-flight tests without
disrupting flight
Health Status Fusion
Flight Control Reconfiguration
System and Component Health
Faults, Health Observable at Component Level
Inform of remaining capability
- Maintenance Support
- Prognostics
- Pilot or Autonomous System Action
- Reconfiguration Compensates for Damage and
Failures
30FSFCC Hardware is Flight Proven
- Developed jointly by NAVAIR and NASA
- Derivative of NASA HARV configuration
- Compatible with any F/A-18A-D
- Requires flight test jumpers installed on MCs
- Requires DAF
- Originally Flight tested at NASA Dryden and at
Patuxent River in 1998 (FSFCC V1.1 Software) - 3 flights at NASA Dryden
- 11 flights at Patuxent River
- All test objectives met
31Pilot Interface
1
2
DDI A to completesequence and arm FSFCC
DDI B, C, D combinations to define test
3
4
For Example DCCBBB Table 22 Row 0 (Fail R Stab
to 0) CCBCB Table 4 Row 3 (Nz Upper Limit
Table Entry 1)
NWS button to engage FSFCC
ADS paddle to disengage FSFCC
32Major Components
- Reference Model (Prescribed Offline)
- A model that encodes the desired aircraft
responses - to pilot inputs as a function of operating
condition, etc. - Low order equivalent system transfer functions
from pilot - stick to aircraft responses (i.e. pitch stick to
pitch rate, - roll stick to roll rate, etc.)
- Model parameters computed from high-fidelity
simulation - data of nominal (unimpaired) aircraft
- Model parameters at a given operating condition
are - functions of input magnitude (see figure)
- Reference model integrated online
- (80 Hz update in the 1750A FSFCC)
- Parameter ID System Model (Adapted Online)
- A model that encodes the dynamical responses of
the aircraft as it maneuvers through the flight
envelope. - State Space model with time varying parameters
- Modified Sequential Least Squares algorithm a
regularized parameter estimation method to
determine system model terms - MSLS update of system model done online (5 Hz
update in the 1750A FSFCC)
1
Example Transfer Function Gain, Roll Axis Ref.
Model
2
33Major Components (contd)
- Model-Based Adaptive Control (Solved Online)
- Online control design procedure that operates on
the system and reference models to generate
control commands that cause aircraft dynamics to
track reference models. - Continuous-time formulation of receding horizon
control for a state-space system model - Optimal solution via differential Riccati
equations replaced with approximate solution to
integrate control law gains directly - 30 percent less memory, 25 percent faster
computation - Closed loop simulation performance comparable for
set of test cases considered - Control gain differential equations solved online
(10 Hz update in 1750A FSFCC) - Most recent control gains applied at basic frame
rate (80 Hz in 1750A FSFCC)
3
34Control Law Cost Function
- RHO is a solution to the finite horizon
optimization problem that minimizes
symmetric positive semidefinite weighting matrix
that assigns importance to predicted tracking
error
symmetric positive semidefinite weighting matrix
that assigns importance to integrated tracking
error
symmetric positive definite weighting matrix that
penalizes control effort