Date post: | 26-Jun-2018 |
Category: |
Documents |
Upload: | hoangkhuong |
View: | 219 times |
Download: | 1 times |
Aviation Safety Program
Integrated Vehicle Health ManagementIntegrated Vehicle Health ManagementTechnical Accomplishments
Principal Investigator: Ashok N. Srivastava, Ph.DProject Scientist: Robert Mah, Ph.D.
Project Manager: Joseph E. Grady, Ph.D.Associate Project Manager: Susan Johnson,
Project Coordinator: Elizabeth FoughtyProject Coordinator: Elizabeth Foughty
www.aeronautics.nasa.govashok@email arc nasa gov
IVHM Review (11/6/07) 1
ARMD ReviewNASA HeadquartersNovember 5, 2007
Key AccomplishmentsHigh Quality Tech Accomplishments
Highly Scalable Prognostics(Inverting Matrices with 10^12 elementsMS 1.8.5, 2.8.3)MS 1.8.5, 2.8.3)
Bayesian Modeling for Diagnostics(Operating in a state space of size 2^400 in
millisecond time, MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
Automatic Identification of Contributing Factors for Aviation Incidents(No supervised training required, MS 3.82, 2.84, 1.8.5)
Contributing Factors Classification
0.6
0.8
1
1.2
NMFMarianaLDA
High Temperature Technologies(3000 Hours of World Record
P f MS 1 3 5)
0
0.2
0.4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Contributing Factor
LDA
0
1
2
3Input Signal (1 V P-P Sinewave)Output after 1 hour @ 500 °COutput after 500 hours @ 500 °C
Sig
nal (
V)
Performance, MS 1.3.5)
Transfer of Distributed National FOQA and ASAP
-2
-1
0 1 2 3 4
S
Time (msec)
IVHM Review (11/6/07) 2
Transfer of Distributed National FOQA and ASAP Archives to FAA(Enabling Fleetwide IVHM through Data Mining, PART Milestone FY07)
Aviation Safety ProgramTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
Bayesian Methods for Diagnosisy g
Ole J. Mengshoel, Ph.D
• Scott Poll Adnan Dar iche
NASA Ames team: UCLA team:• David Nishikawa• Scott Poll
• Ann Patterson-Hine• David Garcia • David Hall
• Adnan Darwiche • Keith Cascio • Mark Chavira• Yuliya Zabiyaka
David Nishikawa• John Ossenfort• Adam Sweet• Serge Yentus
S d U k• Charles Lee• Chris Neukom
Yuliya Zabiyaka• Serdar Uckun
IVHM Review (11/6/07) 3
ARMD ReviewNASA HeadquartersNovember 5, 2007
Integrated Vehicle Health ManagementTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
• Real-time operating system (RTOS) used in current avionics:
Task has: period deadline and worst case– Task has: period, deadline, and worst-case execution time (WCET)
– Priority-based preemptive scheduling• The challenge of embedding AI into a hard real-The challenge of embedding AI into a hard real
time system:– Difficulty of the computational problems– High expectation and/or variance of a search g p
algorithm’s execution time • Embedded IVHM processes for diagnosis,
sensor validation, prognosis: – Should be designed within RTOS resource
bounds – such as WCET– Involve uncertainties – probability of component
and sensor failure
IVHM Review (11/6/07) 4
and sensor failure
Uncertainty and Bayesian ReasoningTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
• Need to represent and reason with uncertain data in IVHM
• Uncertainty reasoning based on probability• Uncertainty reasoning based on probability theory is well-established
• Earlier: Probability theory is elegant, but has serious limitations:
Bayesian Reasoning for Diagnostics: Operates in a state space of size > 2^400
– Requires too much information – joint distribution is exponentially large
– Inefficient and unpredictable algorithms– Is unsuitable for embedded IVHM
state space of size 2 400 in time less than one millisecond.
s u su tab e o e bedded• Recently: Solutions to above issues
– Hardware speed– Bayesian networks (BNs) that decompose the
joint distributionjoint distribution – Efficient algorithms: Compilation of BN to
arithmetic circuit, where computation is fast and predictable
IVHM Review (11/6/07) 5
Our Bayesian Diagnosis ApproachTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
Bayesian Health nodesSensor nodesINPUT:
ObservedOUTPUT: QueryInference
Command nodesea odesObserved
VariablesQuery Variables
432 nodes
• Bayesian network model for a vehicle’s subsystem(s): – It represents health of sensors and
subsystem components explicitlyy y– It contains random variables for
other parts of the subsystem • A Bayesian approach to:
– Diagnosis: health status of systemDiagnosis: health status of system component nodes
– Sensor validation: health status of sensor nodes
IVHM Review (11/6/07) 6
Much simplified but illustrative example
Compilation Approach to DiagnosisTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
Offline Bayesian ababaababaababaababaf |||| θθλλθθλλθθλλθθλλ +++=
Compilationy
Network +
* *
Arithmetic Circuit (AC)
++
* * * * * *
A B
CPTs θ(A)d θ(B|A) Circuit (AC) * *
aλ aλbλ bλaθ aθab|θ ab |θ ab|θ ab |θ
* * * *and θ(B|A)
OnlineInference
(AC Evaluator)
Diagnosis:Belief, MPE,
MAP
Sensor, Commands
IVHM Review (11/6/07) 7
(AC Evaluator)
For details on Arithmetic circuits: [Darwiche, 03; Darwiche & Chavira, 07]
ADAPT TestbedTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
• ADAPT: Electrical power system (EPS) testbed developed at NASA Amestestbed developed at NASA Ames
• EPS loads include: avionics, propulsion, life support, and thermal management pp g
– Increased reliance on EPS in new air- and spacecraft
• ADAPT provides:• ADAPT provides: – a standard testbed for evaluating diagnostic
algorithms & software;bilit f t ll d i ti f f lt– a capability for controlled insertion of faults,
giving repeatable failure scenarios; and – a stepping stone for maturing diagnostic
t h l i
IVHM Review (11/6/07) 8
technologies.
See also http://ti.arc.nasa.gov/projects/adapt/
EPS: Diagnostic ChallengesTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
• Hybrid system with many ibl dpossible modes
• Timing and delay considerations
• Transients
IVHM Review (11/6/07) 9
• Transients• Load-dependent noise
Compilation for Diagnosis in ADAPT Tech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
432 nodes
Offline Compilation
Bayesian Network
SystemTopology
Offline Generation
Arithmetic Circuit (AC)
Component Types
Online
Circuit (AC)
S
yp
OnlineInference
Sensor, Commands
For diagnosis, we perform a probabilistic query over the
IVHM Review (11/6/07) 10
Diagnosis:Belief, MPE
For diagnosis, we perform a probabilistic query over the (currently) 122 health variables in the ADAPT in order to find out which health variable(s) is (are) in non-healthy state(s).
ADAPT Fault Injection ExperimentsTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
#309 #305#309
#311
#305
#310
ADAPT Experiments - Run Times
100 000 #306#308
10.000
100.000
me
(mill
isec
)
MeanMedian
#3040.384 0.471 0.360 0.443 0.340 0.429 0.478
0.100
1.000
304 305 306 308 309 310 311
Run
tim Maximum
IVHM Review (11/6/07) 11
ADAPT Experiment ID
SummaryTech Accomplishments
(IVHM v1.5 MS 1.4.1, 1.4.5, 2.4.3, 3.4.2)
• IVHM challenges: – Complex systems, large state spaces
Uncertainty components and sensors– Uncertainty - components and sensors may fail
– Computationally hard problems – Resource-bounds – real time
Bayesian Reasoning for Diagnostics: Operates in a state space of size > 2^400Resource bounds real time
requirements• Successful application of Bayesian
networks in ADAPT:
state space of size 2 400 in time less than one millisecond.
networks in ADAPT:– Diagnosis and sensor validation under
uncertainty • Benefits of compilation to arithmetic p
circuit:– Speed: 300-400 microseconds (on
average, in software) for ADAPT
IVHM Review (11/6/07) 12
– Predictability: important for real-time IVHM applications
IVHM Technical AccomplishmentTech Accomplishments
(IVHM v1.5 MS 1.3.5)
High Temperature SiC Transistor Integrated Circuits
Philip Neudeck, Gary Hunter, Robert Okojie, Glenn Beheim, Roger Meredith, Terry Ferrier, Laura Evans, Michael Krasowski,
and Norman Prokopand Norman ProkopNASA Glenn Research Center
http://www.grc.nasa.gov/WWW/SiC/
David Spry and Liang-Yu ChenOAI
Dorothy Lukco and Carl ChangASRC
IVHM Review (11/6/07) 13
Aviation Safety Program Review6 November 2007
IVHM - Propulsion Health ManagementTask 3.3: High-Temperature Enabling Technologies
Tech Accomplishments(IVHM v1.5 MS 1.3.5)
“Sensor development will be focused in three primary areas:
Gas PathHealth
1) high-temperature pressure sensors for incorporation into gas-path trending and fault diagnostic models to infer Health
Management IVHMg
turbine health, 2) structural health monitors
including smart accelerometers, and optical , pstrain and blade tip-timing sensors, and
3) high-temperature wireless communications and
StructuralHealth
High TemperatureSensors, Electronics,and Communications
energy harvesting technologies to enable the addition of these high-temperature sensors while
IVHM Review (11/6/07) 14
Management and Communications pminimizing wiring and power requirements on the engine.”
Enabling TechnologyHigh Temperature Integrated Circuitry
Tech Accomplishments(IVHM v1.5 MS 1.3.5)
High temperature semiconductor integrated circuit (IC) chips are the key foundational technology for realizing health management technologies in harsh environments.
Once developed, high-temperature semiconductor IC chips could:- be physically small and light (to put as many as you need, anywhere they are needed).- be highly complex to perform “intelligent” circuit functions.- operate with relatively low power consumption- operate with relatively low power consumption.- operate in hot engine areas without active liquid coolant plumbing/lines.- communicate data wirelessly (to FADEC, or other “distributed control” chips).
CRITICAL REQUIREMENT: High temperature electronic IC’s must operate DURABLY AND RELIABLY in order to be useful to most applications.
Temperatures exceed 300 °C in important areas of engines.Peak compressor gas T near 500 °C, higher for combustor & exhaust.
Above 300 °C ambient, wide bandgap semiconductors (SiC) are needed.- Beyond physical limits of silicon IC’s (including “high T” silicon chips).
- Adverse silicon leakage currents - loss of “semiconductor” behavior.
p g , g
IVHM Review (11/6/07) 15
g- Chemical diffusions and reactions (this is also a problem for GaN).
- Active (liquid) electronics cooling negatively offsets desired system benefits.
High Temperature Semiconductor ElectronicsTech Accomplishments
(IVHM v1.5 MS 1.3.5)
Of the examples of semiconductor transistor integrated circuit demonstrations in published literature, none prior to this work demonstrated stable electrical operation beyond a few hours at T ≥ 500 °C.operation beyond a few hours at T 500 C.
Initial Goal: Highly durable 500-600 °C integrated circuit technology foundation.- Prolonged operation without cooling, physically small, up-scalable.g p g p y y p- Sensor signal conditioning and data transmission (wireless).
NASA’ UNIQUE RANGE OF HARSH ENVIRONMENT
Microfabrication LaboratoryRange of Physical and Chemical Intelligent Sensor
NASA’s UNIQUE RANGE OF HARSH ENVIRONMENT TECHNOLOGY AND CAPABILITIES
Microfabrication LaboratorySiC Clean Room Processing
Range of Physical and Chemical Sensors for Harsh Environments
Intelligent Sensor Mini-Systems
IVHM Review (11/6/07) 16
Previous (to IVHM) Key NASA Glenn AdvancementsTech Accomplishments
(IVHM v1.5 MS 1.3.5)
Key fundamental high temperature electronic materials and processing challenges have been faced and overcome by systematic basic materials processing research (fabrication and characterization)
500 °C Durable Chip PackagingAnd Circuit Boards
( C G C & )
500 °C Durable Metal-SiC Contacts(R. Okojie, 2000 GRC R&T Report)
processing research (fabrication and characterization).
(L. Chen, 2002 GRC R&T Report)
100 um
IVHM Review (11/6/07) 17
Additional advancements in device design, insulator processing, etc. also made.
SiC Transistor Structure6H-SiC Junction Field Effect Transistor (JFET)
Tech Accomplishments(IVHM v1.5 MS 1.3.5)
• 6H p-type SiC wafer with epilayers (purchased from Cree)
• Two p type epilayers
6H-SiC Junction Field Effect Transistor (JFET)
• Two p-type epilayers~1019, ~ 1 µm (buffer layer)~1015, ~ 6-8 µm
• n-type channely1-2x1017, ~0.2-0.4 µm
• p+ gate ~1020, ~ 0.14 µm• Ti/TaSi2/Pt electric contact • Oxide (wet rewet) and nitride• Oxide (wet rewet) and nitride
passivation• Triple-layer contact on the
backside of wafer • TaSi2/Pt interconnect metal
(single layer interconnect) • Metal patterning was dry/wet no
ion damage to dielectric
IVHM Review (11/6/07) 18
ion damage to dielectric
JFET chips enabled by development and integration of new SiC processes.
6H-SiC Junction Field Effect Transistor (JFET)Fabricated by NASA Glenn Research Center
Tech Accomplishments(IVHM v1.5 MS 1.3.5)
200µm/10µm 6H-SiC JFET Packaged with bond wires
200 µm
0 µm
100 µmSource10
Drain Gate
1 mmOptical micrograph of
device before packaging
IVHM Review (11/6/07) 19
Packaged Devices and Test SetupTech Accomplishments
(IVHM v1.5 MS 1.3.5)
1 cm
Parallel fabrication and testing of both single-transistors and IC’s
Boards with chips reside in ovens.
Oxidizing atmospheric air at 500 °C.
IVHM Review (11/6/07) 20
200µm/10µm 6H-SiC JFET
DifferentialAmplifier
Continuous electrical testing at 500 °C.
Wires to test instrumentation.
NASA Glenn SiC JFET : First Transistor to Surpass2000 Hours of Stable Electrical Operation at 500 °C
Tech Accomplishments(IVHM v1.5 MS 1.3.5)
Current-voltage characteristics are very good and stable after 2000 hours.– Enables realization of analog integrated circuits (amplifiers, oscillators).
Excellent turn-off characteristics, large ON to OFF current ratio (> 1000).
5
– Enables realization of digital logic circuits.
Current vs. Voltage Characteristics Key Parameters vs. Time @ 500 °C2.0
O ti Ti t 500 °C
3
4 RDS(ON)
(kΩ)g
m X 10 (mS)I (mA)rV
alue1.5
nt (m
A)
Operating Time at 500 °C1 hour2000 hours
1
2
3 IDSS
(mA)
Par
amet
er
0.5
1.0
ain
Cur
ren
0
1
0 500 1000 1500 2000
P
Time(hours)0.0
0.5
0 10 20 30 40 50
Dra
Drain Voltage (V)
IVHM Review (11/6/07) 21
Time(hours)
Less than 7% change occurs during 2000 hours at 500 °C (most during 1st 100 hrs).- 7% change is smaller than listed on most silicon transistor spec. sheets.
Drain Voltage (V)
NASA Glenn Silicon Carbide Differential AmplifierWorld’s First Semiconductor IC to Surpass
2000 Hours of Electrical Operation at 500 °C
Tech Accomplishments(IVHM v1.5 MS 1.3.5)
p
Demonstrates CRITICAL ability to interconnect transistors and other components (resistors) in a small area on a single SiC chip to form useful integrated circuits that are durable at 500 °C.
2
3Optical micrograph of demonstration amplifier circuit before packaging
Test waveforms at 500 °CInput (1 V P-P Sine Wave)Output 1 hour @ 500 °COutput 2000 hours @ 500 °C
0
1
nal (
V)
Output 2000 hours @ 500 C
-1
0
Sig
n
100 um
-20 1 2
Time (msec)2 transistors and 3 resistors
integrated into less than half a square millimeter
IVHM Review (11/6/07) 22
square millimeter. Less than 3% change inoperating characteristics during
2000 hours of 500 °C operation.Single-metal level interconnect.
NASA Glenn SiC JFET NOR Gate ICWorld’s First Semiconductor Digital IC toSurpass 1000 hours of 500 °C Operation
Tech Accomplishments(IVHM v1.5 MS 1.3.5)
Waveforms of packagedNOR (= “Not OR”) gate at 500
°C
VDDCircuitSchematic
C
IN A OUT A+BIN B - 5 V
0 V
OUT A+B
Time at 500 °C1 hour1270 hours
VSS GND
OUT A B
- 5 V
-10 V 0 V
IN AGND - 5 V
-10 V0 V
IN A
0 5 10 15 20 25 30 35
- 5 V
-10 V
IN B
IVHM Review (11/6/07) 23
Probe-TestPhoto
Time (msec)
Next Steps for SiC IC DevelopmentTech Accomplishments
(IVHM v1.5 MS 1.3.5)
1. Continue high temperature testing.- How long will the chips last at 500 °C?- Analyze failure mechanisms (after circuits finally fail).Analyze failure mechanisms (after circuits finally fail).- Revise process to further increase 500 °C circuit durability.- Test more chips, test at 600 °C (accelerated failure testing).- Test for thermal cycling, thermal shock, vibration, etc.
2. Improve metal interconnects from single-level to double-level process.- Enable much more complicated integrated circuitry.
3. Fabricate improved SiC transistors and prototype integrated circuits.- Higher frequency operation (shrink devices, antennas & passives).- Up-scaled levels of integration (100’s to 1000’s of transistors)Up scaled levels of integration (100 s to 1000 s of transistors).- Complete 500 °C wireless telemetry integrated circuits.- Engine ground test and flight demonstrations.- Demonstrate even higher temperature operation.
IVHM Review (11/6/07) 24
g p p
4. Transfer technology for further development and manufacturing.
SummaryTech Accomplishments
(IVHM v1.5 MS 1.3.5)
NASA has pioneered revolutionary durable 500 °C integrated circuits.New technology foundation -- new applications!
Durable high temperature IC’s will enable important newcapability highly relevant to advancing IVHM.- Enabled by fundamental electronic materials research.
W ld d IC d bilit t 500 °C ( 200 f ld i t)- World record IC durability at 500 °C (> 200-fold improvement).- Approach is inherently up-scalable to high circuit complexity
while remaining physically small.
Further development should lead to beneficial technology insertion.- IC’s for engines (especially new wireless health monitoring).- Technology transfer and commercialization.gy- Additional applications (automotive, well-drilling, Venus).
Chips are still running well at 500 °C.
IVHM Review (11/6/07) 25
p g- Amplifier chip went past 3000 hours on 10/31/07.