Thermography Imaging Based On Equivalent Wave Field Transform For Client
Thermography Imaging Based On Equivalent Wave Field Transform For Client
Based on Equivalent
Wave Field Transform
1
Equivalent Wave Field Transform
• Equivalent Wave Field Transform is unique mathematical algorithm which simplified imaging with
such as thermal propagation.
• By using relationships between wave propagation and thermal diffusive propagation, it was shown
that one can transform from diffusive propagation into an equivalent wave field. The transformation
results in sharper reflections with time delay proportional to the distance.
• US patent 10,089,734 “Equivalent wave field processing of thermal images”.
Dr. Meir Gershenson – Diffusion Research
• A thermal imaging were sample is heated and temperature sequence is recorded using IR camera.
Images are processed to obtain depth information.
• It is a noncontact method.
• Resolving the main holdback of current systems as flaw resolution, computation time, cost and
weight.
2
Equivalent Wave Field Transform
3
Some Math
The thermal heat equation is given by:
𝜕
k∇2 𝑇 𝒓, 𝑡 − 𝜌𝐶 𝑇 𝒓, 𝑡 = ℎ𝑑 𝒓, 𝑡
𝜕𝑡
2 𝜕2 𝜕2 𝜕2
∇ = 2 + + 2 (2)
𝜕𝑥 𝜕𝑦 2 𝜕𝑧
K is the thermal conductivity 𝜌𝐶 the heat capacity and ℎ𝑑 the thermal excitation. The
scalar wave equation is given by:
2 1 𝜕2
∇ 𝑥 𝒓, 𝑡 − 𝑥 𝒓, 𝑡 = ℎ𝑤 𝒓, 𝑡 (3)
𝑣 2 𝜕𝑡 2
The two equations are very similar, the difference is the order of the time derivative,
first order in the heat diffusion equation and second order in the wave equation
4
Some Math
Based on to Morrison at all3, when comparing equation 2 and 3
∞
𝑥𝑑 𝑡 = 0 𝑥𝑤 𝜏 × 𝑊 𝑡, 𝜏 𝑑𝜏 (4a)
𝑒 − 𝑠 𝜏 𝐻𝑑 𝑠
𝑊 𝑡, 𝜏 = ℒ −1 ,𝑡 (4b)
𝐻𝑤 𝑠
Were Hd(s) and Hw(s) are the Laplace transform of hd(t) and hw(t) and ℒ −1 is the inverse Laplace
transform operator. In discreet time domain equation 4 become a matrix multiplication.
𝒙𝒅 = 𝑾 ∙ 𝒙𝑤 (5)
In principle 𝒙𝒅 can be calculated from 𝒙𝑤 by inverting the matrix W, but the condition number of W
is extremely large and the inversion is an ill-posed problem. Ill posed problem is characterized by
high sensitivity to noise, small noise in the input will result in nonphysical big change in the output.
In such a case one uses either truncated singular value decomposition (TSVD) or Tikhonov
regularization to invert the matrix W.
5
Common Analysis Methods
TSR - time domain (proprietary).
Lock-in – Frequency domain (most common).
6
EWFT is a thermal NDT method
• EWFT is a thermal NDT method. It has a great potential to replace current most wide
used Magnetic Particle, Liquid Penetration and Eddy Current Testing methods for many
applications.
11
Pulse Eddy Current Thermography (PECT)
If Eddy Current can see it, Eddy Current Thermography can see it even better 12
Pulse Eddy Current Thermography
Current flow around the crack Early heat distribution cross section
13
System Components
14
Proof of Concept
Weld Sample:
15
Test Results - Lack of Side Wall Fusion
Sequence of images at equal time intervals
16
Test Results - Crack
Sequence of images at equal time intervals
17
Test Results - Crack
Sequence of images at equal time intervals
18
Test Results - Crack
(play short movie)
19
Length and Depth
• The length and depth of
the indications can be
measured.
20
Equivalent Wave Field
Transform Method on
Polycarbonate Sample
21
Modelling
5mm
10mm
Camera - Opgal Therm App Hz, highly reviewed smartphone add on long wave infrared
camera. Properties of the camera are: 384 x 288 pixels, 25Hz frame rate and NETD <0.07°C.
Heat Source - two 500W quartz halogen lamps with Arduino relay driver and a small
photographic studio lamp holder.
A low-end gaming laptop with Linux operating system was used for data collection and
processing
24
Sequence of back projected images at 0.5mm intervals
25
Conclusion
• Equivalent Wave Field Transform is a true 3D method.
• Principal components thermography (PCT) is known for its high resolution. PCT
suffer from few disadvantages, it is extremely computational intensive.
• Equivalent Wave Field Transform analysis results are very similar to PCT in
resolution.
• Equivalent Wave Field Transform is extremely efficient in term of computation, it
requires only single matrix multiplication.
• Taking advantage of the known depth it is possible to apply enhancement
developed for wave field such as delay and sum back projection.
28