Smart Structures Chenyang Lu Cyber-Physical Systems Laboratory - - PowerPoint PPT Presentation
Smart Structures Chenyang Lu Cyber-Physical Systems Laboratory American Society for Civil Engineers 2017 Report Card for America's Infrastructure Bridges C+ q Almost four in 10 are 50 years or older. q 56,007 (9.1%) bridges were structurally
Chenyang Lu
Cyber-Physical Systems Laboratory
American Society for Civil Engineers
Ø Bridges C+
q Almost four in 10 are 50 years or older. q 56,007 (9.1%) bridges were structurally deficient. q Backlog of bridge rehabilitation needs: $123 billion. q https://youtu.be/JjN1FwzbJaY
Ø Dams D Ø Levees D Ø Roads D Ø … … Ø America's Infrastructure GPA: D+
2
http://www.infrastructurereportcard.org
Current Practice Ø Bridges: inspected manually once every two years. Ø Costly and time consuming.
3
Highway 40 Closing for Boone Bridge Inspection Monday August 10, 2009 If you're heading to St. Charles this weekend, Highway 40 is not your best option. Westbound 40 from Long Road in St. Louis County to Route 94 in St. Charles County will be closed (weather permitting) while work crews inspect the Daniel Boone Bridge across the Missouri River. The road will close at 5:30 a.m. on August 15 and won't reopen until sometime after 9 p.m. on August 16.
4
Ø Bridges live for 50-100 years
q Need to make sure they remain
safe for a long time
Ø Smart monitoring and control systems to prevent…
Freeway after 1989 San Francisco Earthquake Minneapolis Bridge Collapse
5
Ø Monitor a bridge using a wireless sensor network
q Detect and localize damages to structures
Ø Smart
q No human effort
Ø Real-time
q Every week q Right after an earthquake
Ø Accurate
q Technology instead of human eyes
Ø Detect and localize damages to structures
q at high spatiotemporal granularities
Ø Challenges
q Computationally intensive q Resource constraints q Long-term monitoring
6
Ø Centralized: stream all data to base station for processing.
q Too energy-consuming for long-term monitoring
Ø Example: Golden Gate Bridge project [Kim IPSN'07].
q Nearly 1 day to collect enough data. q Lifetime of 10 weeks w/4 x 6V lantern battery.
Ø Separate designs of sensor networks (cyber) and damage detection (physical).
Ø Sensor networks focus on data transport. Ø Not concerned with method for damage detection.
7
Ø Dilemma
q Too much sensor data to stream to the base station q Damage detection algorithms are too complex to run
entirely on sensors
➪ Edge computing
q Perform part of computation on sensor nodes q Send (smaller) intermediate results to base station q Complete computation at base station
8
Raw Data
Partial Results
Ø Employ damage detection approach amenable for distributed implementation in sensor networks. Ø Optimally map damage detection algorithm onto distributed architecture.
9
Raw Data
Partial Results
q Physical: Damage Localization Assurance Criterion (DLAC) [Messina96]
q Identify structure’s natural frequencies based on vibration data.
q “Match” natural frequencies to structural models with damages.
q Cyber: optimally partition data flow between sensors and base station.
q Minimize energy consumption q Subject to resource constraints
10
(1) FFT (2) Power Spectrum (3) Curve Fitting (4) DLAC
D Integers Healthy Model Damaged Location 2D Floats D Floats P Floats
D: # of samples P: # of natural freq. (D » P)
Data Flow Analysis DLAC Algorithm
11
(3a) Coefficient Extraction (3b) Equation Solving
5*P Floats
Data Flow Analysis DLAC Algorithm
12
(4) DLAC (1) FFT (2) Power Spectrum (3) Curve Fitting
Healthy Model Damaged Location 8192 bytes 4096 bytes
D: 2048 P: 5 Integer: 2 bytes Float: 4 bytes
(3a) Coefficient Extraction (3b) Equation Solving
100 bytes 20 bytes
Effective compression ratio of 204:1
4096 bytes
Ø Platform: Imote2 + ITS400 sensor board
q 13 – 416 MHz XScale CPU q 32 MB ROM, 32 MB SDRAM q CC2420 802.15.4-compliant radio q 3-axis accelerometer on sensor board
Ø Data collection and processing application written with TinyOS 1.1
q 243 KB ROM, 71 KB RAM
13
Ø 5.6m steel truss structure at UIUC Ø 14 0.4m long bays, on 4 rigid supports Ø 11 Imote2s attached to frontal pane
14
Wireless Sensor
Truss Frontal Panel
123456789 10 11 12 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 X = 3 Y = 0.868 DLAC WS #32 Truss Central Bay Position 123456789 10 11 12 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 X = 3 Y = 0.864 DLAC WS #45 Truss Central Bay Position 123456789 10 11 12 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 X = 3 Y = 0.871 DLAC WS #67 Truss Central Bay Position 123456789 10 11 12 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 X = 3 Y = 0.873 DLAC WS #28 Truss Central Bay Position 123456789 10 11 12 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 X = 3 Y = 0.825 DLAC WS #35 Truss Central Bay Position 123456789 10 11 12 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 X = 3 Y = 0.865 DLAC WS #75 Truss Central Bay PositionDamage correctly localized to third bay
Energy Consumption
0.05 0.1 0.15 0.2 0.25 Decentralized Centralized Energy consumption (mAh) Sampling Computation Communication
Evaluation
15
Energy Consumption
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Raw Data Collection FFT Power Spectrum Coefficient Extraction Equation Solving Energy Consumption (mAh) Sampling Computation Communication
Evaluation
16
Memory Consumption
50000 100000 150000 200000 250000 RAM ROM Size (bytes) Raw Data Collection FFT Power Spectrum Coefficient Extraction Equation Solving
Evaluation
17
Ø Cyber-physical co-design of a distributed SHM system.
q Reduces energy consumption by 71% q Implemented on iMote2 using <1% of its memory
Ø Effectively localized damage on two physical structures. Ø Demonstrated the promise of cyber-physical co-design.
18
Structural Damage Localization Using Wireless Sensor Networks, RTSS 2008.
Ø The DLAC method employs no collaboration among sensors à limitations in SHM capabilities.
q For example, cannot detect multiple damages.
Ø New hierarchical architecture for collaborative localization.
q Embed processing into a hierarchical architecture q Send (smaller!) partial results between layers of hierarchy q Multi-level damage localization
Ø Demonstrate the generality of cyber-physical co-design
19
Ø Flexibility acts as a “signature” of the structure’s health
q Structures flex slightly when a force is applied q Structural weakening => decreased stiffness
Ø Two flexibility-based methods of interest for our work
q Beam-like structures: Angles-Between-String-and-Horizon flexibility method
(ASHFM) [Duan, J. Structural Engineering and Mechanics 09]
q Truss-like structures: Axial Strain flexibility method (ASFM) [Yan, J. Smart Structures
and Systems 09]
20
θ
Ø Sensors form groups Ø Group members
q collect raw vibration data q à power spectrum
Ø Group leaders
q collect and correlate power
spectrum from children
q à modal parameters (natural
frequencies + mode shapes)
21
Base Station Group Leader Group Leader Group Member Group Member Group Member Group Member Group Member
Ø Base station
q collects modal parameters from group leaders q à structural flexibility q compared against “baseline” collected when
structure was known to be healthy
Ø Differences in flexibility à localize damage
22
23
Sensing FFT Power Spectrum Cross Spectral Density Singular Value Decomposition
2D ints D floats
Group Leader
Flexibility
Base Station
D matrices
Group Member
D floats P natural frequencies + mode shapes D: # of samples P: # of natural freq. (D » P)
24
Sensing FFT Power Spectrum
2D ints D floats
Peak Picking
D floats
Cross Spectral Density Singular Value Decomposition
Group Leader
Flexibility
Base Station
P matrices P natural frequencies + mode shapes P floats
Group Member
D: # of samples P: # of natural freq. (D » P)
Ø Only a handful of sensors are needed to detect damage Ø As more sensors are added, localization gets more precise Ø Save energy by exploiting localized nature of flexibility-based approach
25
Ø Only a handful of sensors are needed to detect damage Ø As more sensors are added, localization gets more precise Ø Save energy by exploiting localized nature of flexibility-based approach
26
Ø Only a handful of sensors are needed to detect damage Ø As more sensors are added, localization gets more precise Ø Save energy by exploiting localized nature of flexibility-based approach
27
Ø Hardware: Imote2 + ITS400 sensor board
q 13 – 416 MHz PXA271 XScale CPU q 32 MB ROM, 32 MB SDRAM q CC2420 802.15.4-compliant radio q 3-axis accelerometer on sensor board
Ø Software platform
q TinyOS 1.1 operating system q UIUC’s ISHM toolsuite used for sensing,
reliable communication, and time sync
28
Ø Simulation of 5.6 m, 14-member steel truss structure at UIUC Ø Simulated sensor data generated in MATLAB and injected into live application using “fake” sensor driver
q Intact data set: no damages q Damaged data set: three members reduced on left side of truss,
four on right side
29
Ø Level 1: nine sensors at uniform points along truss’s length
30
2 4 6 8 1 2 3 4x 10
−6
Element Number Damage Indicator
Damage identified
Damage identified
Ø Level 2: move all nine sensors to respective halves (emulate higher density)
31
1 2 3 4 5 6 7 8 9 10111213 1 2 3 Element Number Damage Indicator 1 2 3 4 5 6 7 8 9 10111213 1 2 3 Element Number Damage Indicator
Damage localized correctly to all seven members
Ø Codesigned architecture reduces communication latency from estimated 87s to 0.21s Ø 78.9% of energy attributable to synchronization and sensing Ø Compare to theoretical energy supply of 20,250 J (3x 1.5 V, 1250 mAh AAA batteries)
32
Group Member Synchronization 12.1 J Sensing 23.0 J Computation 9.28 J Communication 0.08 J Group Leader Synchronization 16.2 J Sensing 21.2 J Computation 8.52 J Communication 0.76 J
Full-Scale Truss
33
Image source: Zhuoxiong Sun, Purdue University
Ø Two levels of damage localization Ø Level 1: localized damage to bay 9 Ø Level 2: localized damage to element 42
34
2 3 4 5 6 7 8 910 20 31 32 42 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 x 10
−6Truss Element Number AS Flexibility Damage Indicator
Ø Cyber-physical co-design for wireless structural health monitoring
q Distribute flexibility-based damage localization methods in a
hierarchical architecture
q Multi-level search strategy only activates sensors in area of
interest; many sensors remain asleep
Ø Localize damage to a simulated truss and a real full-size truss with low energy consumption Ø Long-term goal: a general cyber-physical co-design approach to integrated sensing and control
35
Distributed Structural Health Monitoring with Wireless Sensor Networks, IEEE Transactions
Ø Localize damages on structures using wireless sensors. Ø Traditional: separate network and civil engineering
q Cyber: Wireless network streams all data to a base station q Physical: Base station runs damage localization algorithm
Ø Clean separation of concern, but ineffective
q Streaming raw data consumes too much energy
Ø Get hands dirty Ø Understand the data flow of damage localization Ø But still employ clean abstraction and methodology Ø Optimal data flow embedding in a network Ø Highly effective Ø Reduces energy consumption by 71% [RTSS'08]
37
Ø Structural control systems protect civil infrastructure. Ø Wired control systems are costly and fragile. Ø Wireless structural control achieves flexibility and low cost.
4/2/18 38
Heritage tower crumbles down in earthquake of Finale Emilia, Italy, 2012. Hanshin Expressway Bridge after Kobe earthquake, Japan, 1995.
Ø Wireless Cyber-Physical Simulator (WCPS)
q Capture dynamics of both physical plants and wireless networks q Enable holistic, high-fidelity simulation of wireless control systems q Integrate TOSSIM and Simulink/MATLAB q Open source: http://wcps.cse.wustl.edu
Ø Realistic case studies on wireless structural control
q Wireless traces from real-world environments q Structural models of a building and a large bridge q Excited by real earthquake signal traces
Ø Cyber-physical co-design
q End-to-end scheduling + control design q Improve control performance under wireless delay and loss
4/2/18 39
Ø Bill Emerson Memorial Bridge
q Main span: 1,150 ft. q Carries up to 14,000 cars a day
q New Madrid Seismic Zone
Ø Replace joints by 24 hydraulic actuators Ø Vibration mode:
q 0.1618 Hz for 1st mode q 0.2666 Hz for 2nd mode q 0.3723 Hz for 3rd mode
4/2/18 40
(a) (b)
Ø Largest wireless bride deployment [Jang 2010]
q 113 Imote2 units; Peak acceleration sensitivity of 5mg – 30mg
Ø RSSI/noise traces from 58-node deck-network for this study
4/2/18 41
42
Cyber-physical co-design à 50% reduction in control power.
43