Industrial Panel Meeting

1
Industrial Panel Meeting May 13 2008 PG Group
Research Portfolio Patrick Gaydecki and Bosco
Fernandes School of Electrical and Electronic
Engineering University of Manchester PO Box
88 Manchester M60 1QD United Kingdom Tel
UK-44 (0) 161 306 4906 patrick.gaydecki_at_manches
ter.ac.uk www.eee.manchester.ac.uk/research/gro
ups/sisp/research/dsp
2
Inductive Scan Imaging
3

• Inductive scan imaging systems generate images of
  embedded steel by analysing the response of a
  time-varying magnetic field that is impressed on
  the material by resonant transmitter coils.
• Since 1996 the team has published over 40 papers
  to journals and conference proceedings
• By exploiting both the changes in impedance and
  inductance of the coil, corrosion can be imaged.
• The team is now developing a system to detect
  corrosion in the pre-stressing wire of concrete
  pipes.

4
filter/ gain
peak detector
ripple filter
precision offset null
tuned oscillator
Sensing Coil
reference oscillator
phase detector
frequency to voltage converter
precision offset null
filter/ gain
DAC
ADC
real-time DSP system
The Imaging Sensor
processor for filtering and control
local memory
comms port
5
Early Experimental Configuration
6
Image from Sensor bar diameter 16mm, upper bar
depth 20mm scan height 20mm above surface
7
Determination of Scan Profile Parameters
8
Curve Fitting Approach

• Fit a curve to the sensor response (Pearson VII
  function).
• Extract the peak parameters peak value and full
  width at half height.
• Use a curve-fitting model or train a neural
  network on the extracted peak parameters to
  estimate the bar dimensional information.

9
3-D Bar Visualisation
10
3x3 bar mesh using 16mm bar size. The top layer
is at depth of 30 mm.
11
Dimensional Result 16mm Bars, Scan Depth 30mm
Calculated Items Top Layer Bottom Layer
?1 (v1, w1) 16.40 16.61
?2 (v2, w2) 16.39 16.54
?3 (v3, w3) 16.65 16.54
d1 (v1, ? 16) 29.65 45.70
d2 (v2, ? 16) 29.75 46.01
d3 (v3, ? 16) 30.26 45.42
d1 (v1, w1) 29.77 46.26
d2 (v2, w2) 29.88 46.52
d3 (v3, w3) 30.55 45.92
12
Image Result 16mm Bars, Scan Depth 30mm
13
Imaging of Steel Bars Located behind a Ferrous
Steel Layer
14
3 Coil Configuration
function generator
amplification
subtraction
power amplifier

• receiver

• transmitter

• dummy

15
Coil Arrangement
16
Bars placed under a 0.5mm thick mild steel plate
17
Broken bar placed under a 2 mm thick mild steel
plate
18
Corrosion Detection, Imaging and Quantification
19
Corrosion Quantification Experimental Setup
3x2 samples from each bar size
solution concentration of 0.2
The bar samples
The accelerated corrosion system
20
Corrosion Quantification Initial Results
21
Image of steel bar and of corrosion product
(Q-detection)
22
Image of corrosion product and steel
bar (phase-detection)
23
Image from Heterodyne output bar diameter 20mm,
scan height 30mm above surface
24
Qualitative corrosion estimation bar diameter
20mm, bar length 200mm, scan height 25mm
25
The Solid State Magnetic Field Camera (mFIC)
26
System Configuration

• Previous studies have shown that a 2D array with
  a minimum 33 x 33 sensor elements will provide
  sufficient information to allow the generation of
  a high resolution images using image
  interpolation techniques.
• It has also been shown that in a 3D spatial
  orientation, the vertical component of the
  magnetic flux density (Bz) is the most favourable
  for image generation, since the image is easy to
  interpret and can readily be applied to
  reconstruct the objects geometry.
• In these experiments we have deployed a 1D linear
  array of 33 fix solid-state sensors oriented
  along the Bz axis, in conjunction with a 300 x 40
  mm rectangular coil, to produce line scan
  readings similar to those obtained previously in
  a traditional 2D scan comprising a single sensor.
• A third generation mFIC has also been
  constructed, which comprises a 2D array of 33 x
  33 sensor elements, which scans electronically
  and has no moving parts (discussed at the end of
  this report). However, it was deemed that the
  system was not optimised for the present
  feasibility tests and was therefore not used.

27
Linear Array (1D) System Design

• The PCB sensor array was constructed using 33 MI
  solid-state sensors (PNI Corp. PNI Sen-s65)
  spaced with a pitch of 9.325 mm and aligned in a
  sensor probe that was vertically mounted in order
  to measure the magnetic flux density along the Z
  axis (Bz).
• Each sensor has a total field range from 11 to
  11 gauss, with a typical resolution of
  approximately 0.015 µT.
• The readings from each sensor were acquired and
  digitised sequentially by using an Application
  Specific Integrated Circuit (ASIC) module,
  controlled from a central DSP microprocessor
  unit. The DSP unit also coordinated the
  multiplexing and data acquisition via an SPI
  interface.
• Data were stored in memory during the scan and
  finally transmitted to a computer via an RS 232
  interface.

28
Linear Array Design
ASIC control and excitation circuit
DSP data controller
DC current system
29
Linear Array Design
Version 1.1
Version 2.1
30
Experimental Setup
31
Experimental Setup
32
Image Pre-Processing

• The raw image generation process proceeds as
  follows
• An initial scan is taken without the target and
  with the excitation current switched on, to
  calibrate the sensors.
• A second scan is taken of the target with the
  excitation current switched off, to estimate the
  targets residual magnetic field.
• A final scan is taken of the target with the
  excitation current switched on.
• The final scan is corrected for variations in the
  sensor response and the residual magnetic field
  of the target.

Scan 1 No target, field on
Scan 2 With target, field off
Scan 3 With target, field on
Compensation
Pre-processed Image
33
Image Post-Processing
Multistage image processing algorithm to separate
the different layers and to de-blur the
composite image
Benitez et al., Efficient image enhancement
algorithm for images of steel reinforcing bars in
concrete obtained by a new solid-state
sensor-based system, IET Science, Measurement
Technology , Volume 1, Issue 5, p. 255-260, 2007.
34
Validation of Modelling Optimum Number of Sensors
Experimental outcomes confirmed simulated
predictions
35
Image Processing Results
Scanner images bar mesh located at 100 mm depth
36
Image Processing Results Images of 12mm bar mesh
at 100 mm depth
original
original
original
processed
processed
processed
37
Results Concrete Block Scanning
Rebar configuration within the block
Scanned image
38
Complete Linear Array Scanner System
39
Scan of Target under Ceramic Tile and Foil
40
Scan of Hammer under Plasterboard
41
Scan of Hammer under Plasterboard and Foil
42
Test Matrix
Target / Barrier Air Foil in air Ceramic Tile Tile and foil Plaster board Plaster board and foil
Wire strippers Yes Yes Yes Yes Yes Yes
Pliers Yes No No No Yes Yes
Hammer Yes Yes Yes Yes Yes Yes
Kitchen knife No No Yes Yes Yes Yes
Steel disc Yes Yes Yes Yes Yes Yes
Notes All plasterboard 9.5 mm thickness All
ceramic tiles 8 mm thickness
43
Metal Object Imaging in Air, 25 mm
44
Results Images of Wire-Strippers
45
Results Images of Pliers
46
(No Transcript)
47
Sample Post-Processed Image
48
Magnetic Field Imaging of a Ring Coil with DC
Current Excitation
117 mm depth
49
2D mFIC Configuration

• A solid state mFIC has now been fabricated which
  comprises an array of 33 x 33 magneto-inductive
  sensors and a square DC excitation coil.
• Three sub-controllers, each of which is
  responsible for handling the data from eleven
  linear arrays, feed data to a master controller,
  which in turn communicates with the host
  computer.
• Scanning is performed entirely electronically and
  involves no moving parts.
• An image is generated within three seconds
  future developments will enable a data rate of
  five frames (images) per second.
• At present the device is optimised for imaging
  steel bars embedded in concrete. Modifications of
  the system will be required to enable it to
  detect and image metal weaponry concealed behind
  walls.
• However, it should be noted that a modified
  linear mFIC, scanned manually, may be more
  appropriate for the imaging of concealed weaponry.

50
2D Array Design
DSP Sub-controller 1
DSP Main Controller
DSP Sub-controller 2
DSP Sub-controller 3
51
Completed Solid State 2D Array Scanner
52
Development of a Large Bore Prestressed Pipe
Inspection System
53
Hardware
Digital Function generator
Power amplifier
Transmitter coil
Instrumentation amplifier
DC power supply
DSP phase/amplitude system
Digital filter system
Instrumentation amplifier
Computer
Oscilloscope
Receiver coil
54
Early Rotational Line-scan System
Pipe testing
55
Rotational Line-scan Result
Tx over broken wires
Rx over broken wires
56
Brega Plant Tests
57
Brega Plant 1.6 m Pipe Rotational Scan
Rx over breaks
Tx over breaks
58
Brega Plant 4m Pipe Wire Break Spot Tests
59
Brega Plant 4m Pipe Line Scan Tests
60
Digital Signal Processing and its Application
to Low Level Signal Recovery
61
Characteristics of DSP Systems

• DSP offers flexibility, allowing a single
  platform to be rapidly reconfigured for different
  applications
• Operations such as modulation, phase shifting,
  signal mixing and delaying are simply performed
  in software
• System performance is far more accurate than
  equivalent analogue systems
• However, for real time operation, considerable
  intellectual investment is required to design and
  program DSP platforms

62
DSP Hardware and Software Specifications (Signal
Wizard Systems, Developed at UoM)

• 24-bit codec resolution (1 part in 16777216, or
  144 dB)
• Variable sample rate extending to 196 kHz
• Processor power of 100 - 550 million
  multiplication-accumulations per second (MMACS)
• Dual channel/ eight channel operation for signal
  referencing and mixing with digital audio
  inputs/outputs
• Non-volatile memory for retention of settings
• Standard PC interfacing USB, JTAG, serial and
  parallel
• Easy FIR, and IIR filter design, with other
  processing functions including mixing and phasing

63
Final Realized Hardware (Signal Wizard 2)
64
Signal Wizard 3
550 million multiplications and additions per
second
65
Signal Wizard 2 Software
FIR and IIR design area
Graphical display of filter
Hardware control download, gain, adaptive,
delay, mixing etc.
66
Adaptive Filter Software
67
Signal Shape Reconstruction Essential Equations
Describing Time Domain Deconvolution (Inverse
Filtering) for Real-Time and Off-Line Processing
Simple Fourier domain scheme (rarely successful)
Fourier domain scheme with noise estimate
Time domain scheme with noise estimate
(surprisingly useful)
Finally
68
Signal Shape Reconstruction in Practice
Signal comprising three impulses
Signal after inverse filtering in real-time
Signal after low-pass distortion
69
Detection of AC magnetic fields propagated
through a ferrous steel boundary The Skin Effect
The strength of both the eddy currents and the
associated magnetic field fall rapidly with depth
in ferrous materials. The equation which
describes the fall in current density is given by
Where Js is the current density on the surface, d
is the depth within the material, d is the skin
depth, w is the angular frequency and J is the is
the current density at depth d. The skin depth
for a given material is governed by the
relationship
where s is the conductivity of the conductor or
target, mr its relative permeability and m0 is
the absolute permeability of a vacuum.
70
Experimental Configuration
Steel properties Relative permeability 250 Cond
uctivity 6.0 ? 106 Sm-1
Frequency, kHz Transmitted flux density, T Skin depth, mm Attenuation Received flux density, T
4.5 3.1 ? 10-4 0.194 3.33 ? 10-5 1.03 ? 10-8
9.0 1.73 ? 10-4 0.137 4.57 ? 10-7 7.89 ? 10-11
13.0 4.09 ? 10-4 0.114 2.20 ? 10-8 8.99 ? 10-12
Digital Oscilloscope
Instrumentation amplifier (x 400)
Mild steel plate
Mild steel enclosure
digital function generator
power amplifier
71
Laboratory System
72
IIR Filter Frequency Response at 9 kHz
Pole locations for two IIR filters Pole locations for two IIR filters Pole locations for two IIR filters Pole locations for two IIR filters Pole locations for two IIR filters
Centre Frequency, kHz p0, real p0, imaginary p1, real p1, imaginary
4.5 0.83146 0.55556 0.83146 -0.55556
9.0 0.38268 0.92387 0.38268 -0.92387
73
Line Scan Results at 4.5 kHz and 9 kHz
4.5 kHz
9 kHz
74
Detection of low amplitude ultrasonic pulses
propagated through seawater via a steel structure
7 m structure being lowered into the dock at
Liverpool, UK
Location of transmitter
75
Experimental Configuration
Signal type Tone burst
Frequency 40 kHz
Transmission amplitude 20 V
Divergence angle hemispherical
Attenuation Geometric (1/r2)
Peak received signal 650 nV
Digital Oscilloscope
Instrumentation amplifier (x 400)
Microcontrolled pulser
100 m
Receiving transducer
Transmitting transducer
76
Typical Results
(a)
(c)

1. Detail of original received signal degraded by
   noise.
2. Detail of received signal, recovered by super
   narrowband filter.
3. Complete tone burst signal detected after
   transmission through water, recovered using a
   super narrowband IIR filter.

(b)
77
Intelligent Clothing Development
78
Sensor Electronics
Analogue electronics
Digitization
BT wireless Tx
79
Data Acquisition and DSP
HR detection
DSP
ECG
Channel splitting
BT wireless Rx
COM PORT
volume display
DSP
respiration
80
Standard medical equipment and the SmartLife
system A comparative study
Standard ECG methods have been compared against
the SmartLife system in the following ways

• Comparison of the responses from standard
  Silver/Silver Chloride (Ag/AgCl) electrodes and
  yarn electrodes using hospital monitoring
  equipment
• Comparison of the above responses using the
  Smartlife electronics and software system
• Analysis of the data from Holter Monitors and
  Loop Recorders

81
Comparative study Results
Standard hospital equipment
Ag/AgCl
Vest
Smartlife electronics and software
Holter monitor
Loop recorder
82
Signal comparison yarn and standard electrodes
Signal from standard Ag/AgCl Gel electrodes
Signal from SmartLife Health Vest electrodes
Signal Section Amplitude (mV) Amplitude (mV) Duration (ms) Duration (ms)
Signal Section Ag/AgCl Vest Ag/AgCl vest
P wave 0.2 0.3 120 120
QRS complex 2.0 2.5 80 80
T wave 0.5 0.5 240 240
83
(No Transcript)