SlideShare a Scribd company logo
Seoul	National	University
Wigner-Ville Distribution:
In Perspective of Fault Diagnosis
(Based on Time-Frequency Analysis, Cohen and
Time-Frequency Toolbox for Use with Matlab,
Auger)
Jungho Park, Ph.D Candidate
System Health & Risk Management Laboratory
Department of Mechanical & Aerospace Engineering
Seoul National University
Seoul	National	University2018/1/27 - 2 -
Contents
4.	Second	class	of	solutions:	the	energy	distribution
4.1.	The	Cohen’s	class
4.1.1.	The	Wigner-Ville	distribution
4.1.2.	The	Cohen’s	class
4.1.3.	Link	with	the	narrow-band	ambiguity	function
4.1.4.	Other	important	energy	distribution
4.1.5.	Conclusion
Time-Frequency Toolbox
For Use with MATLAB
8.	The	Wigner	Distribution
9.	General	Approach	and	the	Kernel	Method
10.	Characteristic	Function	Operator	Method
11.	Kernel	Design	for	Reduced	Interference
12.	Some	Distributions
Time-Frequency Analysis,
Cohen
Seoul	National	University
• First	class	of	solutions:	Atomic	
decomposition
• Fourier	transform
• Short-time	Fourier	transform
• Wavelet	transform
2018/1/27 - 3 -
8. The Wigner Distribution
• Definition	(Related	to	the	energy of	the	signals)
• Second	class	of	solutions:	Energy	
distribution
• Wigner	Distribution
• Choi-Williams	distribution
• Zhao-Atlas-Marks
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
𝑋 𝜈 = ' 𝑥 𝑡 𝑒01234: 𝑑𝑡
78
08
𝐹 𝑥 𝑡, 𝜈; ℎ = ' 𝑥 𝑢 ℎ∗(𝑢 − 𝑡)𝑒01234: 𝑑𝑢
78
08
𝑇 𝑥 𝑡, 𝑎; Ψ = ' 𝑥 𝑠 Ψ:,C
∗
(𝑠)𝑑𝑠
78
08
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗(𝑡 −
𝜏
2
)𝑒012345 𝑑𝜏
78
08
𝑃 𝐶𝑊 𝑡, 𝜔 =
1
4𝜋J/2
' '
1
𝜏2/𝜎
exp[−
(𝑢 − 𝑡)2
4𝜏2/𝜎
− 𝑗𝜏𝜔]
×𝑠∗
𝑢 − 𝜏/2 ℎ 𝑢 + 𝜏/2 𝑑𝑢𝑑𝜏
𝑍𝐴𝑀 𝑥 𝑡, 𝑣 = ' ℎ(𝜏) ' 𝑥 𝑠 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)	𝑑𝑠
:7 5 /2
:0 5 /2
𝑒012345
𝑑𝜏
78
08
Seoul	National	University
• Property	(Refer	to	Cohen	to	check	the	proof)
1. Real	value
• The	calculated	values	are	real
(It	can	be	proved	by	the	fact	that	the	distribution	and	its	complex	
conjugate	are	same.)
2018/1/27 - 4 -
• Definition	(Related	to	the	energy of	the	signals)
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
𝑊∗
𝑡, 𝜔 =
1
2𝜋
' 𝑠 𝑡 +
𝜏
2
𝑠∗
(𝑡 −
𝜏
2
)𝑒15Z
𝑑𝜏
= −
[
23
∫ 𝑠 𝑡 +
5
2
𝑠∗
(𝑡 −
5
2
)𝑒015Z
𝑑𝜏
08
8
=
[
23
∫ 𝑠 𝑡 +
5
2
𝑠∗
(𝑡 −
5
2
)𝑒015Z
𝑑𝜏
8
08
																									= 𝑊(𝑡, 𝜔)
Seoul	National	University
𝐸 = ' ' 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡 = ' 𝑠(𝑡) 2 𝑑𝜏 = 1
• Property	(Refer	to	Cohen	to	check	the	proof)
2. Marginality
• The	energy	spectral	density	 𝑺(𝝎) 𝟐 and	the	instantaneous	power	 𝒔(𝒕) 𝟐 can	be	
obtained	by	marginal	distribution	of	the	Wigner	distribution
2018/1/27 - 5 -
• Definition	(Related	to	the	energy of	the	signals)
Wigner	distribution	is	energy	distribution
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
𝑃 𝑡 = ' 𝑊 𝑡, 𝜔 𝑑𝜔 =
1
2𝜋
' ' 𝑠∗
𝑡 −
𝜏
2
𝑠 𝑡 +
𝜏
2
𝑒015Z
𝑑𝜏𝑑𝜔
= ∫ 𝑠∗
𝑡 −
5
2
𝑠 𝑡 +
5
2
𝛿(𝜏)𝑑𝜏
= 𝑠(𝑡) 2
Seoul	National	University
• Property	(Refer	to	Cohen	to	check	the	proof)
3. Non-positivity
• The	distribution	could	have	negative	values	(Contradictory	to	the	concept	of	energy	
density)
2018/1/27 - 6 -
• Definition	(Related	to	the	energy of	the	signals)
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
(Figure	from	Cohen)
Seoul	National	University2018/1/27 - 7 -
• How	the	negative	values	are	treated	in	the	literature
Normal 50%	fault
100%	fault
Staszewski, Wieslaw J., Keith Worden, and Geof R. Tomlinson.
"Time–frequency analysis in gearbox fault detection using the
Wigner–Ville distribution and pattern recognition." Mechanical
systems and signal processing 11.5 (1997): 673-692. 327 cited
“The	negative	values	of	the	distribution	
were	set	to	zero	to	avoid	difficulties	with	
the	physical	interpretation.”
Baydar, Naim, and Andrew Ball. "A comparative study of
acoustic and vibration signals in detection of gear failures
using Wigner–Ville distribution." Mechanical systems and
signal processing 15.6 (2001): 1091-1107. 272 cited
Normal
25%	fault
50%	fault
“To	overcome	this	problem	and	reduce	the	presence	
of	interference	components,	a	smoothed	version	of	
the	WVD	(SPWVD)	is	used.”
8. The Wigner Distribution
Seoul	National	University
• Property	(Refer	to	Cohen	to	check	the	proof)
4. Global	average
2018/1/27 - 8 -
• Definition	(Related	to	the	energy of	the	signals)
ß Global	average
(due	to	marginal	property)
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
< 𝑔[ 𝑡 + 𝑔2 𝜔 >= ' ' 𝑔[ 𝑡 + 𝑔2 𝜔 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡
= ∫ 𝑔[ 𝑡 𝑠(𝑡) 2
𝑑𝑡 + ∫ 𝑔2 𝜔 𝑆(𝜔) 2
𝑑𝜔
< 𝑔 𝑡, 𝜔 >= ' ' 𝑔 𝑡, 𝜔 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡
Seoul	National	University
• Property	(Refer	to	Cohen	to	check	the	proof)
5. Local	average
• Instantaneous	frequency	and	group	delay can	be	derived	from	local	averages	of	the	
Wigner	distribution
2018/1/27 - 9 -
• Definition	(Related	to	the	energy of	the	signals)
ß Local	average
𝜑 :	phase
𝜓 :	spectral	phase
Instantaneous	frequency Group	delay
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
< 𝜔 >:=
1
𝑠(𝑡) 2
' 𝜔𝑊 𝑡, 𝜔 𝑑𝜔 < 𝑡 >Z=
1
𝑆(𝜔) 2
' 𝑡𝑊 𝑡, 𝜔 𝑑𝑡
𝑡
;
< 𝜔 >:= 𝜑′(𝑡) ; < 𝑡 >Z= −𝜓′(𝜔)
Seoul	National	University
• Property	(Refer	to	Cohen	to	check	the	proof)
6. Time	and	Frequency	shift
2018/1/27 - 10 -
• Definition	(Related	to	the	energy of	the	signals)
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
if			𝑠 𝑡 → 𝑒1Zn:
𝑠 𝑡 − 𝑡o 			then			𝑊 𝑡, 𝜔 → 𝑊(𝑡 − 𝑡o,𝜔 − 𝜔o)
𝑊st 𝑡, 𝜔 =
1
2𝜋
' 𝑒01Zn :05/2
𝑠∗
(𝑡 − 𝑡o −
𝜏
2
)	
×𝑒1Zn :75/2
𝑠(𝑡 − 𝑡o +
5
2
)𝑒015Z
𝑑𝜏
=
[
23
∫ 𝑠∗
(𝑡 − 𝑡o −
5
2
)𝑠(𝑡 − 𝑡o +
5
2
) 𝑒015(Z0Zn)
𝑑𝜏
= 𝑊(𝑡 − 𝑡o, 𝜔 − 𝜔o)
Seoul	National	University
• Property (Refer	to	Cohen	to	check	the	proof)
7. Cross-term (Interference)
• For	multi-component signals,	cross-terms come	out	due	to	quadratic	calculation
2018/1/27 - 11 -
• Definition	(Related	to	the	energy of	the	signals)
Cross-terms
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
𝑠 𝑡 =𝑠1 𝑡 +𝑠2 𝑡
𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 𝑊12 𝑡, 𝜔 + 𝑊21 𝑡, 𝜔
where	𝑊12 𝑡, 𝜔 = ' 𝑠[
∗
	 𝑡 −
𝜏
2
𝑠2(𝑡 +
𝜏
2
)𝑒015Z 𝑑𝜏
𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 2Re	{𝑊12 𝑡, 𝜔 }
(Figure	from	Auger)
Seoul	National	University2018/1/27 - 12 -
• Definition	(Related	to	the	energy of	the	signals)
8. The Wigner Distribution
• Property (Refer	to	Cohen	to	check	the	proof)
7. Cross-term (Interference)
• For	multi-component signals,	cross-terms come	out	due	to	quadratic	calculation
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
Cross-terms
𝑠 𝑡 =𝑠1 𝑡 +𝑠2 𝑡
𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 𝑊12 𝑡, 𝜔 + 𝑊21 𝑡, 𝜔
where	𝑊12 𝑡, 𝜔 = ' 𝑠[
∗
	 𝑡 −
𝜏
2
𝑠2(𝑡 +
𝜏
2
)𝑒015Z 𝑑𝜏
𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 2Re	{𝑊12 𝑡, 𝜔 }
(Figure	from	Cohen)
Seoul	National	University2018/1/27 - 13 -
• Definition	(Related	to	the	energy of	the	signals)
ü First	let	us	make	clear	that	it	is	not	generally	
true	that	the	cross	terms	produce	undesirable	
effects.	~~~	In	fact,	since	any	signal	can	be	
broken	up	into	a	sum	of	parts	in	an	arbitrary	
way,	the	cross	terms	can	be	neither	bad	nor	
good	since	they	are	not	uniquely	defined;	they	
are	different	for	different	decompositions.	The	
Wigner	distribution	does	not	know	about	cross	
terms,	since	the	breaking	up	of	a	signal	into	
parts	is	not	unique.	(P.126,	Cohen)
ü However,	the	localization	and	amplitude	of	
these	additional	terms	often	make	the	use	
and	interpretation	of	the	representation	
difficult,	or	even	impossible when	the	signal	
contains	a	large	number	of	“elementary	
components”.	Since	these	interference	terms	
distribute	the	real	part	of	the	scalar	product	in	
the	time-frequency	plane,	they	distribute	
negative	values	when	the	scalar	product	is	
negative.	 (P.	148-149,	Auger)
8. The Wigner Distribution
• Property (Refer	to	Cohen	to	check	the	proof)
7. Cross-term (Interference)
• Two	difference	views	on	cross-terms
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
Seoul	National	University
• Property
• Instantaneous	frequency and	group	
delay can	be	derived	by	local	average.
• The	outputs	could	have	negative	values,	
which	is	counter-intuitive.
• Suffers	from	the	fact	that	confusing	
artifacts	could	be	achieved	for	
multicomponent	signals	(Cross-terms)
2018/1/27 - 14 -
• Comparison	between	the	Wigner	distribution	and	the	spectrogram
Wigner	distribution Spectrogram
• Property
• Instantaneous	frequency and	group	
delay can	only	be	approximated.
• The	outputs	always	have	positive	
values.
• The	multi-component	could	not	be	
effectively	resolved.	(Window	size)
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗(𝑡 −
𝜏
2
)𝑒012345 𝑑𝜏
78
08
𝐹 𝑥 𝑡, 𝜈; ℎ = ' 𝑥 𝑢 ℎ∗(𝑢 − 𝑡)𝑒01234z 𝑑𝑢
78
08
Seoul	National	University2018/1/27 - 15 -
• Smoothed-pseudo	Wigner-Ville	distribution	(SPWVD):	To	solve	cross-term	problems
WVD:
PWVD:
SPWVD:
(Smoothing	in	frequency-domain)
(Smoothing	both	in	time- and	frequency-domain)
8. The Wigner Distribution
𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
𝑃𝑊 𝑥 𝑡, 𝜈 = ' ℎ(𝜏)𝑥 𝑡 +
𝜏
2
𝑥∗
(𝑡 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
𝑆𝑃𝑊 𝑥 𝑡, 𝜈 = ' ℎ(𝜏) ' 𝑔(𝑠 − 𝑡)𝑥 𝑠 +
𝜏
2
𝑥∗
(𝑠 −
𝜏
2
)𝑒012345
𝑑𝜏
78
08
78
08
Seoul	National	University2018/1/27 - 16 -
• Smoothed-pseudo	Wigner-Ville	distribution	(SPWVD):	To	solve	cross-term	problems
(figure	from	Auger)
WVD PWVD SPWVD
Smoothing	in	freq. Smoothing	in	time
8. The Wigner Distribution
Seoul	National	University2018/1/27 - 17 -
• Definition
(Cohen)
(Auger)
Kernel	function
Parameterization	function
• Types	of	kernels
• Product	kernel:	General	case
• Separable	kernel
9.	General	Approach	and	the	Kernel	Method	(The	Cohen’s	class)
𝐶 𝑡, 𝜔 =
1
4𝜋2
' ' ' 𝑠∗
𝑢 −
𝜏
2
𝑠 𝑢 +
𝜏
2
𝜙 𝜃, 𝜏 𝑒01}:015Z71}z
𝑑𝑢𝑑𝜏𝑑𝜃
𝐶~ 𝑡, 𝜐; 𝑓 = ' ' ' 𝑒123• s0: 𝑓(𝜉, 𝜏)𝑥 𝑠 +
𝜏
2
𝑥∗(𝑠 −
𝜏
2
)𝑒012345 𝑑𝜉𝑑𝑠𝑑𝜏
78
08
𝜙(𝜃, 𝜏) = 𝜙ƒ„ 𝜃𝜏 = 𝜙(𝜃𝜏)
𝜙 𝜃, 𝜏 = 𝜙[(𝜃)𝜙[(𝜏)
Seoul	National	University2018/1/27 - 18 -
• Some	Distributions	and	Their	Kernels
(Table	from	Cohen)
9.	General	Approach	and	the	Kernel	Method	(The	Cohen’s	class)
Seoul	National	University2018/1/27 - 19 -
• Basic	properties	related	to	the	kernel
• Marginals:	Instantaneous	Energy	/	
Energy	Density	Spectrum
Basic	form
Integrating	wrt
frequency
For	the	
integration	to	be	
instantaneous	
power	
( )For	frequency	marginal
For	total	energy
9.	General	Approach	and	the	Kernel	Method	(The	Cohen’s	class)
𝐸 = ' ' 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡 = ' 𝑠(𝑡) 2 𝑑𝜏
𝑃 𝑡 = ' 𝑊 𝑡, 𝜔 𝑑𝜔 = 𝑠(𝑡) 2
𝐶 𝑡, 𝜔 =
1
4𝜋2
' ' ' 𝑠∗
𝑢 −
𝜏
2
𝑠 𝑢 +
𝜏
2
𝜙 𝜃, 𝜏 𝑒01}:015Z71}z
𝑑𝑢𝑑𝜏𝑑𝜃
'𝐶 𝑡, 𝜔 𝑑𝜔 =
1
2𝜋
' ' ' 𝛿(𝜏)𝑠∗
𝑢 −
𝜏
2
𝑠 𝑢 +
𝜏
2
𝜙 𝜃, 𝜏 𝑒1}(z0:)	
𝑑𝑢𝑑𝜏𝑑𝜃
=
1
2𝜋
' '𝜙 𝜃, 0 𝑠(𝑢) 2
𝑒1}(z0:)	
𝑑𝜃𝑑𝑢
1
2𝜋
'𝜙 𝜃, 0 𝑒1}(z0:)	
𝑑𝜃 = 𝛿(𝑡 − 𝑢)
𝜙 𝜃, 0 =1
𝜙 0, 𝜏 =1
𝜙 0,0 =1
Seoul	National	University2018/1/27 - 20 -
• Basic	properties	related	to	the	kernel
• Time	and	frequency	shift
• Scaling	invariance
• Local	average
• Global	average
• …
9.	General	Approach	and	the	Kernel	Method	(The	Cohen’s	class)
𝐶st 𝑡, 𝜔 =
1
4𝜋2	
' ' ' 𝑒01Zn(z0
5
2
0:n)	
𝑒1Zn(z7
5
2
0:n)	
×				𝑠∗
𝑢 −
5
2
− 𝑡o 𝑠 𝑢 +
5
2
− 𝑡o 𝜙 𝜃, 𝜏 𝑒01}:015Z71}z
𝑑𝑢𝑑𝜏𝑑𝜃
=
1
4𝜋2	
' ' ' 𝜙 𝜃, 𝜏 𝑠∗
𝑢 −
𝜏
2
𝑠 𝑢 +
𝜏
2
𝑒01}:015(Z0Zn)71}(z7:n)
𝑑𝑢𝑑𝜏𝑑𝜃
=
1
4𝜋2	
' ' ' 𝜙 𝜃, 𝜏 𝑠∗
𝑢 −
𝜏
2
𝑠 𝑢 +
𝜏
2
𝑒01}(:0:n)015(Z0Zn)71}z
𝑑𝑢𝑑𝜏𝑑𝜃
= 𝐶 𝑡 − 𝑡o, 𝜔 − 𝜔o
Seoul	National	University2018/1/27 - 21 -
• Objective:	To	maintain	the	good	properties	of	the	Wigner	distribution
11.	Kernel	Design	for	Reduced	Interference
where
*Weak	finite	support
*Strong	finite	support
For	product	kernel,	𝜙(𝜃, 𝜏) = 𝜙ƒ„ 𝜃𝜏 = 𝜙(𝜃𝜏)
(Table	from	Cohen)
ℎ 𝑡 =
1
2𝜋
'𝜙 𝑥 𝑒1~: 𝑑𝑥							; 						𝜙 𝜃𝜏 = 'ℎ 𝑡 𝑒01}5: 𝑑𝑡
𝑃 𝑡, 𝜔 		= 				0				for	𝑡	outside	 𝑡[, 𝑡2 		if	𝑠 𝑡 	is	zero	outside 𝑡[, 𝑡2
𝑃 𝑡, 𝜔 		= 				0				for	𝜔	outside	 𝜔[, 𝜔2 		if	𝑆 𝜔 	is	zero	outside 𝜔[, 𝜔2
𝑃 𝑡, 𝜔 		= 				0				if	𝑠 𝑡 = 0	for	a	particular	time
𝑃 𝑡, 𝜔 		= 				0				if	𝑆 𝜔 = 0	for	a	particular	frequency
Seoul	National	University2018/1/27 - 22 -
• Choi-Williams	method
• Properties
• Product	kernel
• Both	marginal	are	satisfied	(The	energy	spectral	density	 𝑺(𝝎) 𝟐 and	the	
instantaneous	power	 𝒔(𝒕) 𝟐 can	be	obtained)	
• Distribution
12.	Some	distributions
*H.I.	Choi:	Faculty	of	the	Global	School	Of	Media	at	the	Soongsil University
*W.J.	Williams:	Faculty	of	the	Department	of	Electrical	Engineering	and	Computer	Science	at	the	University	of	Michigan	
(For	frequency	marginal) (For	time	marginal)
Kernel	function
! ", $ =
1
4() * ** +∗
- −
/
2
+ - +
/
2
2 3, / 45678569:;67<
=-=/=3
𝜙 𝜃, 𝜏 = 𝑒0}‘5‘/’
𝜙 0, 𝜏 = 1 𝜙 𝜃, 0 = 1
𝑃“” 𝑡, 𝜔 =
1
4𝜋J/2	
' '
1
𝜏2	/𝜎	
exp
(𝑢 − 𝑡)2
4𝜏2/𝜎
− 𝑗𝜏𝜔
	
×				𝑠∗
𝑢 −
5
2
𝑠 𝑢 +
5
2
𝑑𝑢𝑑𝜏
Seoul	National	University2018/1/27 - 23 -
• Choi-Williams	method:	Examples
• For	the	sum	of	two	sine	waves	(																																									),
the	distribution	will	be	calculated	as	
where
à The	distribution	would	have	a	large	peak	at	𝝎 =
𝝎 𝟏
7𝝎 𝟐
𝟐
for	large	𝝈
12.	Some	distributions
Wigner	distribution C-W	with	a	large	𝝈 C-W	with	a	small	𝝈
*C-W	becomes	WD	for	𝜎 → 	∞
𝜙 𝜃, 𝜏 = 𝑒0}‘5‘/’
𝑠 𝑡 = 𝐴[ 𝑒1Z˜:
+ 𝐴2 𝑒1Z‘:
𝐶“” 𝑡, 𝜔 = 𝐴[
2
𝛿 𝜔 − 𝜔[ + 𝐴2
2
𝛿 𝜔 − 𝜔2 + 2𝐴[ 𝐴2 cos[ 𝜔2 − 𝜔[ 𝑡]𝜂(𝜔, 𝜔[, 𝜔2, 𝜎)
𝜂 𝜔, 𝜔[, 𝜔2, 𝜎 =
1
4𝜋 𝜔[ − 𝜔2
2/𝜎
exp
𝜔 −
1
2
	𝜔[ + 𝜔2
2
4𝜋 𝜔[ − 𝜔2
2/𝜎
Figure	from	Cohen
Seoul	National	University2018/1/27 - 24 -
• Choi-Williams	method
12.	Some	distributions
WD
C-W
Spectrogram
Figure	from	Cohen
Seoul	National	University2018/1/27 - 25 -
• Born-Jordan	Distribution:	Reduced	interference
• Zhao-Atlas-Marks	Distribution:	Reduced	interference	by	placing	cross-terms	
under	the	self-terms
12.	Some	distributions
𝜙 𝜃, 𝜏 =
sin(𝑎𝜃𝜏)
𝑎𝜃𝜏
𝜙š›œ 𝜃, 𝜏 = 𝑔 𝜏 𝜏
sin(𝑎𝜃𝜏)
𝑎𝜃𝜏
Figure	from	Cohen
Seoul	National	University2018/1/27 - 26 -
Literature	review
Feng, Zhipeng, Ming Liang, and Fulei Chu. "Recent advances in time–frequency analysis methods for machinery fault diagnosis:
A review with application examples." Mechanical Systems and Signal Processing 38.1 (2013): 165-205. 283 cited
• Linear	time–frequency	representation
STFT WT
Signal: 	𝑥 𝑡 = sin 2𝜋𝑓 ¡¢£ 𝑡 + 2 cos 2𝜋𝑓¤¥¦¦¡£¦ 𝑡 + 153.6 cos 2𝜋𝑓«¬ 𝑡 + 𝑛(𝑡)
Seoul	National	University2018/1/27 - 27 -
Literature	review
Feng, Zhipeng, Ming Liang, and Fulei Chu. "Recent advances in time–frequency analysis methods for machinery fault diagnosis:
A review with application examples." Mechanical Systems and Signal Processing 38.1 (2013): 165-205. 283 cited
• Bilinear	time–frequency	distribution
WVD SPWVD C-H
Signal: 	𝑥 𝑡 = sin 2𝜋𝑓 ¡¢£ 𝑡 + 2 cos 2𝜋𝑓¤¥¦¦¡£¦ 𝑡 + 153.6 cos 2𝜋𝑓«¬ 𝑡 + 𝑛(𝑡)
Seoul	National	University2018/1/27 - 28 -
Literature	review
• Basic	principles	of	gear	fault	diagnosis	à Based	on	side-band	detection
*Feng, Zhipeng, and Ming Liang. "Fault diagnosis of wind turbine planetary gearbox under nonstationary conditions via adaptive
optimal kernel time–frequency analysis." Renewable Energy 66 (2014): 468-477. 56 cited
* *
The	interference	terms	from	WVD	would	make	it	difficult	to	diagnose	the	fault	in	the	system
Seoul	National	University
THANK YOU
FOR LISTENING
2018/1/27 - 29 -

More Related Content

What's hot

RF Circuit Design - [Ch1-2] Transmission Line Theory
RF Circuit Design - [Ch1-2] Transmission Line TheoryRF Circuit Design - [Ch1-2] Transmission Line Theory
RF Circuit Design - [Ch1-2] Transmission Line Theory
Simen Li
 
Discrete fourier transform
Discrete fourier transformDiscrete fourier transform
Discrete fourier transform
MOHAMMAD AKRAM
 
Multiband Transceivers - [Chapter 4] Design Parameters of Wireless Radios
Multiband Transceivers - [Chapter 4] Design Parameters of Wireless RadiosMultiband Transceivers - [Chapter 4] Design Parameters of Wireless Radios
Multiband Transceivers - [Chapter 4] Design Parameters of Wireless Radios
Simen Li
 
1.introduction to signals
1.introduction to signals1.introduction to signals
1.introduction to signals
INDIAN NAVY
 
Chap04
Chap04Chap04
Ch1
Ch1Ch1
Lecture No:1 Signals & Systems
Lecture No:1 Signals & SystemsLecture No:1 Signals & Systems
Lecture No:1 Signals & Systems
rbatec
 
Lecture5 Signal and Systems
Lecture5 Signal and SystemsLecture5 Signal and Systems
Lecture5 Signal and Systems
babak danyal
 
L3. Decision Trees
L3. Decision TreesL3. Decision Trees
L3. Decision Trees
Machine Learning Valencia
 
Crc
CrcCrc
Chapter5 - The Discrete-Time Fourier Transform
Chapter5 - The Discrete-Time Fourier TransformChapter5 - The Discrete-Time Fourier Transform
Chapter5 - The Discrete-Time Fourier Transform
Attaporn Ninsuwan
 
Linear Predictive Coding
Linear Predictive CodingLinear Predictive Coding
Linear Predictive Coding
Shruti Bhatnagar Dasgupta
 
Lecture 4: Classification of system
Lecture 4: Classification of system Lecture 4: Classification of system
Lecture 4: Classification of system
Jawaher Abdulwahab Fadhil
 
Signal modelling
Signal modellingSignal modelling
Signal modelling
Debangi_G
 
Introduction to multiple signal classifier (music)
Introduction to multiple signal classifier (music)Introduction to multiple signal classifier (music)
Introduction to multiple signal classifier (music)
Milkessa Negeri
 
Finite difference method
Finite difference methodFinite difference method
Finite difference method
Divyansh Verma
 
RF Module Design - [Chapter 1] From Basics to RF Transceivers
RF Module Design - [Chapter 1] From Basics to RF TransceiversRF Module Design - [Chapter 1] From Basics to RF Transceivers
RF Module Design - [Chapter 1] From Basics to RF Transceivers
Simen Li
 
Time domain specifications of second order system
Time domain specifications of second order systemTime domain specifications of second order system
Time domain specifications of second order system
Syed Saeed
 
Signals and systems-3
Signals and systems-3Signals and systems-3
Signals and systems-3
sarun soman
 
DSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and Systems
DSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and SystemsDSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and Systems
DSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and Systems
Amr E. Mohamed
 

What's hot (20)

RF Circuit Design - [Ch1-2] Transmission Line Theory
RF Circuit Design - [Ch1-2] Transmission Line TheoryRF Circuit Design - [Ch1-2] Transmission Line Theory
RF Circuit Design - [Ch1-2] Transmission Line Theory
 
Discrete fourier transform
Discrete fourier transformDiscrete fourier transform
Discrete fourier transform
 
Multiband Transceivers - [Chapter 4] Design Parameters of Wireless Radios
Multiband Transceivers - [Chapter 4] Design Parameters of Wireless RadiosMultiband Transceivers - [Chapter 4] Design Parameters of Wireless Radios
Multiband Transceivers - [Chapter 4] Design Parameters of Wireless Radios
 
1.introduction to signals
1.introduction to signals1.introduction to signals
1.introduction to signals
 
Chap04
Chap04Chap04
Chap04
 
Ch1
Ch1Ch1
Ch1
 
Lecture No:1 Signals & Systems
Lecture No:1 Signals & SystemsLecture No:1 Signals & Systems
Lecture No:1 Signals & Systems
 
Lecture5 Signal and Systems
Lecture5 Signal and SystemsLecture5 Signal and Systems
Lecture5 Signal and Systems
 
L3. Decision Trees
L3. Decision TreesL3. Decision Trees
L3. Decision Trees
 
Crc
CrcCrc
Crc
 
Chapter5 - The Discrete-Time Fourier Transform
Chapter5 - The Discrete-Time Fourier TransformChapter5 - The Discrete-Time Fourier Transform
Chapter5 - The Discrete-Time Fourier Transform
 
Linear Predictive Coding
Linear Predictive CodingLinear Predictive Coding
Linear Predictive Coding
 
Lecture 4: Classification of system
Lecture 4: Classification of system Lecture 4: Classification of system
Lecture 4: Classification of system
 
Signal modelling
Signal modellingSignal modelling
Signal modelling
 
Introduction to multiple signal classifier (music)
Introduction to multiple signal classifier (music)Introduction to multiple signal classifier (music)
Introduction to multiple signal classifier (music)
 
Finite difference method
Finite difference methodFinite difference method
Finite difference method
 
RF Module Design - [Chapter 1] From Basics to RF Transceivers
RF Module Design - [Chapter 1] From Basics to RF TransceiversRF Module Design - [Chapter 1] From Basics to RF Transceivers
RF Module Design - [Chapter 1] From Basics to RF Transceivers
 
Time domain specifications of second order system
Time domain specifications of second order systemTime domain specifications of second order system
Time domain specifications of second order system
 
Signals and systems-3
Signals and systems-3Signals and systems-3
Signals and systems-3
 
DSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and Systems
DSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and SystemsDSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and Systems
DSP_2018_FOEHU - Lec 03 - Discrete-Time Signals and Systems
 

Similar to Wigner-Ville Distribution: In Perspective of Fault Diagnosis

Fault detection of a planetary gear under variable speed conditions
Fault detection of a planetary gear under variable speed conditionsFault detection of a planetary gear under variable speed conditions
Fault detection of a planetary gear under variable speed conditions
Jungho Park
 
Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...
Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...
Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...
m.a.kirn
 
Indoor Heading Estimation
 Indoor Heading Estimation Indoor Heading Estimation
Indoor Heading Estimation
Alwin Poulose
 
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
IOSR Journals
 
D012122229
D012122229D012122229
D012122229
IOSR Journals
 
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
IOSR Journals
 
Intelligent fault diagnosis for power distribution systemcomparative studies
Intelligent fault diagnosis for power distribution systemcomparative studiesIntelligent fault diagnosis for power distribution systemcomparative studies
Intelligent fault diagnosis for power distribution systemcomparative studies
nooriasukmaningtyas
 
Joe Kelleher Presentation (May 27th 2014)
Joe Kelleher Presentation (May 27th 2014)Joe Kelleher Presentation (May 27th 2014)
Joe Kelleher Presentation (May 27th 2014)
Roadshow2014
 
Galerkin’s indirect variational method in elastic stability analysis of all e...
Galerkin’s indirect variational method in elastic stability analysis of all e...Galerkin’s indirect variational method in elastic stability analysis of all e...
Galerkin’s indirect variational method in elastic stability analysis of all e...
eSAT Publishing House
 
Calculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolversCalculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolvers
QunaSys
 
Calculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolversCalculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolvers
TenninYan
 
NNBAR SESAPS PRESENTATION FINAL
NNBAR SESAPS PRESENTATION FINALNNBAR SESAPS PRESENTATION FINAL
NNBAR SESAPS PRESENTATION FINAL
Joshua Barrow
 
22. 23767.pdf
22. 23767.pdf22. 23767.pdf
22. 23767.pdf
TELKOMNIKA JOURNAL
 
Detection of crack location and depth in a cantilever beam by vibration measu...
Detection of crack location and depth in a cantilever beam by vibration measu...Detection of crack location and depth in a cantilever beam by vibration measu...
Detection of crack location and depth in a cantilever beam by vibration measu...
eSAT Journals
 
Syllabus
SyllabusSyllabus
Syllabus
rahees1116
 
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORINGSEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
cscpconf
 
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORINGSEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
csandit
 
[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)
[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)
[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)
Hemant Solanki
 
BP219 class 4 04 2011
BP219 class 4 04 2011BP219 class 4 04 2011
BP219 class 4 04 2011
waddling
 
Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...
Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...
Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...
Shu Tanaka
 

Similar to Wigner-Ville Distribution: In Perspective of Fault Diagnosis (20)

Fault detection of a planetary gear under variable speed conditions
Fault detection of a planetary gear under variable speed conditionsFault detection of a planetary gear under variable speed conditions
Fault detection of a planetary gear under variable speed conditions
 
Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...
Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...
Using Machine Learning to Measure the Cross Section of Top Quark Pairs in the...
 
Indoor Heading Estimation
 Indoor Heading Estimation Indoor Heading Estimation
Indoor Heading Estimation
 
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
 
D012122229
D012122229D012122229
D012122229
 
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
Crack Detection for Various Loading Conditions in Beam Using Hilbert – Huang ...
 
Intelligent fault diagnosis for power distribution systemcomparative studies
Intelligent fault diagnosis for power distribution systemcomparative studiesIntelligent fault diagnosis for power distribution systemcomparative studies
Intelligent fault diagnosis for power distribution systemcomparative studies
 
Joe Kelleher Presentation (May 27th 2014)
Joe Kelleher Presentation (May 27th 2014)Joe Kelleher Presentation (May 27th 2014)
Joe Kelleher Presentation (May 27th 2014)
 
Galerkin’s indirect variational method in elastic stability analysis of all e...
Galerkin’s indirect variational method in elastic stability analysis of all e...Galerkin’s indirect variational method in elastic stability analysis of all e...
Galerkin’s indirect variational method in elastic stability analysis of all e...
 
Calculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolversCalculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolvers
 
Calculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolversCalculating transition amplitudes by variational quantum eigensolvers
Calculating transition amplitudes by variational quantum eigensolvers
 
NNBAR SESAPS PRESENTATION FINAL
NNBAR SESAPS PRESENTATION FINALNNBAR SESAPS PRESENTATION FINAL
NNBAR SESAPS PRESENTATION FINAL
 
22. 23767.pdf
22. 23767.pdf22. 23767.pdf
22. 23767.pdf
 
Detection of crack location and depth in a cantilever beam by vibration measu...
Detection of crack location and depth in a cantilever beam by vibration measu...Detection of crack location and depth in a cantilever beam by vibration measu...
Detection of crack location and depth in a cantilever beam by vibration measu...
 
Syllabus
SyllabusSyllabus
Syllabus
 
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORINGSEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
 
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORINGSEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
SEQUENTIAL CLUSTERING-BASED EVENT DETECTION FOR NONINTRUSIVE LOAD MONITORING
 
[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)
[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)
[Ajaya kumar gupta]_response_spectrum_method_in_se(book_zz.org)
 
BP219 class 4 04 2011
BP219 class 4 04 2011BP219 class 4 04 2011
BP219 class 4 04 2011
 
Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...
Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...
Network-Growth Rule Dependence of Fractal Dimension of Percolation Cluster on...
 

Recently uploaded

smart pill dispenser is designed to improve medication adherence and safety f...
smart pill dispenser is designed to improve medication adherence and safety f...smart pill dispenser is designed to improve medication adherence and safety f...
smart pill dispenser is designed to improve medication adherence and safety f...
um7474492
 
原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样
原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样
原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样
ydzowc
 
Zener Diode and its V-I Characteristics and Applications
Zener Diode and its V-I Characteristics and ApplicationsZener Diode and its V-I Characteristics and Applications
Zener Diode and its V-I Characteristics and Applications
Shiny Christobel
 
一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理
一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理
一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理
upoux
 
Ericsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.pptEricsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.ppt
wafawafa52
 
FULL STACK PROGRAMMING - Both Front End and Back End
FULL STACK PROGRAMMING - Both Front End and Back EndFULL STACK PROGRAMMING - Both Front End and Back End
FULL STACK PROGRAMMING - Both Front End and Back End
PreethaV16
 
AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...
AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...
AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...
Paris Salesforce Developer Group
 
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
ijseajournal
 
SCALING OF MOS CIRCUITS m .pptx
SCALING OF MOS CIRCUITS m                 .pptxSCALING OF MOS CIRCUITS m                 .pptx
SCALING OF MOS CIRCUITS m .pptx
harshapolam10
 
P5 Working Drawings.pdf floor plan, civil
P5 Working Drawings.pdf floor plan, civilP5 Working Drawings.pdf floor plan, civil
P5 Working Drawings.pdf floor plan, civil
AnasAhmadNoor
 
一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理
一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理
一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理
upoux
 
一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理
uqyfuc
 
SENTIMENT ANALYSIS ON PPT AND Project template_.pptx
SENTIMENT ANALYSIS ON PPT AND Project template_.pptxSENTIMENT ANALYSIS ON PPT AND Project template_.pptx
SENTIMENT ANALYSIS ON PPT AND Project template_.pptx
b0754201
 
Determination of Equivalent Circuit parameters and performance characteristic...
Determination of Equivalent Circuit parameters and performance characteristic...Determination of Equivalent Circuit parameters and performance characteristic...
Determination of Equivalent Circuit parameters and performance characteristic...
pvpriya2
 
一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理
一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理
一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理
nedcocy
 
Null Bangalore | Pentesters Approach to AWS IAM
Null Bangalore | Pentesters Approach to AWS IAMNull Bangalore | Pentesters Approach to AWS IAM
Null Bangalore | Pentesters Approach to AWS IAM
Divyanshu
 
Object Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOADObject Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOAD
PreethaV16
 
Unit -II Spectroscopy - EC I B.Tech.pdf
Unit -II Spectroscopy - EC  I B.Tech.pdfUnit -II Spectroscopy - EC  I B.Tech.pdf
Unit -II Spectroscopy - EC I B.Tech.pdf
TeluguBadi
 
Digital Twins Computer Networking Paper Presentation.pptx
Digital Twins Computer Networking Paper Presentation.pptxDigital Twins Computer Networking Paper Presentation.pptx
Digital Twins Computer Networking Paper Presentation.pptx
aryanpankaj78
 
Accident detection system project report.pdf
Accident detection system project report.pdfAccident detection system project report.pdf
Accident detection system project report.pdf
Kamal Acharya
 

Recently uploaded (20)

smart pill dispenser is designed to improve medication adherence and safety f...
smart pill dispenser is designed to improve medication adherence and safety f...smart pill dispenser is designed to improve medication adherence and safety f...
smart pill dispenser is designed to improve medication adherence and safety f...
 
原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样
原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样
原版制作(Humboldt毕业证书)柏林大学毕业证学位证一模一样
 
Zener Diode and its V-I Characteristics and Applications
Zener Diode and its V-I Characteristics and ApplicationsZener Diode and its V-I Characteristics and Applications
Zener Diode and its V-I Characteristics and Applications
 
一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理
一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理
一比一原版(uofo毕业证书)美国俄勒冈大学毕业证如何办理
 
Ericsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.pptEricsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.ppt
 
FULL STACK PROGRAMMING - Both Front End and Back End
FULL STACK PROGRAMMING - Both Front End and Back EndFULL STACK PROGRAMMING - Both Front End and Back End
FULL STACK PROGRAMMING - Both Front End and Back End
 
AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...
AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...
AI + Data Community Tour - Build the Next Generation of Apps with the Einstei...
 
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
 
SCALING OF MOS CIRCUITS m .pptx
SCALING OF MOS CIRCUITS m                 .pptxSCALING OF MOS CIRCUITS m                 .pptx
SCALING OF MOS CIRCUITS m .pptx
 
P5 Working Drawings.pdf floor plan, civil
P5 Working Drawings.pdf floor plan, civilP5 Working Drawings.pdf floor plan, civil
P5 Working Drawings.pdf floor plan, civil
 
一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理
一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理
一比一原版(osu毕业证书)美国俄勒冈州立大学毕业证如何办理
 
一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理一比一原版(USF毕业证)旧金山大学毕业证如何办理
一比一原版(USF毕业证)旧金山大学毕业证如何办理
 
SENTIMENT ANALYSIS ON PPT AND Project template_.pptx
SENTIMENT ANALYSIS ON PPT AND Project template_.pptxSENTIMENT ANALYSIS ON PPT AND Project template_.pptx
SENTIMENT ANALYSIS ON PPT AND Project template_.pptx
 
Determination of Equivalent Circuit parameters and performance characteristic...
Determination of Equivalent Circuit parameters and performance characteristic...Determination of Equivalent Circuit parameters and performance characteristic...
Determination of Equivalent Circuit parameters and performance characteristic...
 
一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理
一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理
一比一原版(爱大毕业证书)爱荷华大学毕业证如何办理
 
Null Bangalore | Pentesters Approach to AWS IAM
Null Bangalore | Pentesters Approach to AWS IAMNull Bangalore | Pentesters Approach to AWS IAM
Null Bangalore | Pentesters Approach to AWS IAM
 
Object Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOADObject Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOAD
 
Unit -II Spectroscopy - EC I B.Tech.pdf
Unit -II Spectroscopy - EC  I B.Tech.pdfUnit -II Spectroscopy - EC  I B.Tech.pdf
Unit -II Spectroscopy - EC I B.Tech.pdf
 
Digital Twins Computer Networking Paper Presentation.pptx
Digital Twins Computer Networking Paper Presentation.pptxDigital Twins Computer Networking Paper Presentation.pptx
Digital Twins Computer Networking Paper Presentation.pptx
 
Accident detection system project report.pdf
Accident detection system project report.pdfAccident detection system project report.pdf
Accident detection system project report.pdf
 

Wigner-Ville Distribution: In Perspective of Fault Diagnosis

  • 1. Seoul National University Wigner-Ville Distribution: In Perspective of Fault Diagnosis (Based on Time-Frequency Analysis, Cohen and Time-Frequency Toolbox for Use with Matlab, Auger) Jungho Park, Ph.D Candidate System Health & Risk Management Laboratory Department of Mechanical & Aerospace Engineering Seoul National University
  • 2. Seoul National University2018/1/27 - 2 - Contents 4. Second class of solutions: the energy distribution 4.1. The Cohen’s class 4.1.1. The Wigner-Ville distribution 4.1.2. The Cohen’s class 4.1.3. Link with the narrow-band ambiguity function 4.1.4. Other important energy distribution 4.1.5. Conclusion Time-Frequency Toolbox For Use with MATLAB 8. The Wigner Distribution 9. General Approach and the Kernel Method 10. Characteristic Function Operator Method 11. Kernel Design for Reduced Interference 12. Some Distributions Time-Frequency Analysis, Cohen
  • 3. Seoul National University • First class of solutions: Atomic decomposition • Fourier transform • Short-time Fourier transform • Wavelet transform 2018/1/27 - 3 - 8. The Wigner Distribution • Definition (Related to the energy of the signals) • Second class of solutions: Energy distribution • Wigner Distribution • Choi-Williams distribution • Zhao-Atlas-Marks 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝑋 𝜈 = ' 𝑥 𝑡 𝑒01234: 𝑑𝑡 78 08 𝐹 𝑥 𝑡, 𝜈; ℎ = ' 𝑥 𝑢 ℎ∗(𝑢 − 𝑡)𝑒01234: 𝑑𝑢 78 08 𝑇 𝑥 𝑡, 𝑎; Ψ = ' 𝑥 𝑠 Ψ:,C ∗ (𝑠)𝑑𝑠 78 08 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗(𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝑃 𝐶𝑊 𝑡, 𝜔 = 1 4𝜋J/2 ' ' 1 𝜏2/𝜎 exp[− (𝑢 − 𝑡)2 4𝜏2/𝜎 − 𝑗𝜏𝜔] ×𝑠∗ 𝑢 − 𝜏/2 ℎ 𝑢 + 𝜏/2 𝑑𝑢𝑑𝜏 𝑍𝐴𝑀 𝑥 𝑡, 𝑣 = ' ℎ(𝜏) ' 𝑥 𝑠 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 ) 𝑑𝑠 :7 5 /2 :0 5 /2 𝑒012345 𝑑𝜏 78 08
  • 4. Seoul National University • Property (Refer to Cohen to check the proof) 1. Real value • The calculated values are real (It can be proved by the fact that the distribution and its complex conjugate are same.) 2018/1/27 - 4 - • Definition (Related to the energy of the signals) 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝑊∗ 𝑡, 𝜔 = 1 2𝜋 ' 𝑠 𝑡 + 𝜏 2 𝑠∗ (𝑡 − 𝜏 2 )𝑒15Z 𝑑𝜏 = − [ 23 ∫ 𝑠 𝑡 + 5 2 𝑠∗ (𝑡 − 5 2 )𝑒015Z 𝑑𝜏 08 8 = [ 23 ∫ 𝑠 𝑡 + 5 2 𝑠∗ (𝑡 − 5 2 )𝑒015Z 𝑑𝜏 8 08 = 𝑊(𝑡, 𝜔)
  • 5. Seoul National University 𝐸 = ' ' 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡 = ' 𝑠(𝑡) 2 𝑑𝜏 = 1 • Property (Refer to Cohen to check the proof) 2. Marginality • The energy spectral density 𝑺(𝝎) 𝟐 and the instantaneous power 𝒔(𝒕) 𝟐 can be obtained by marginal distribution of the Wigner distribution 2018/1/27 - 5 - • Definition (Related to the energy of the signals) Wigner distribution is energy distribution 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝑃 𝑡 = ' 𝑊 𝑡, 𝜔 𝑑𝜔 = 1 2𝜋 ' ' 𝑠∗ 𝑡 − 𝜏 2 𝑠 𝑡 + 𝜏 2 𝑒015Z 𝑑𝜏𝑑𝜔 = ∫ 𝑠∗ 𝑡 − 5 2 𝑠 𝑡 + 5 2 𝛿(𝜏)𝑑𝜏 = 𝑠(𝑡) 2
  • 6. Seoul National University • Property (Refer to Cohen to check the proof) 3. Non-positivity • The distribution could have negative values (Contradictory to the concept of energy density) 2018/1/27 - 6 - • Definition (Related to the energy of the signals) 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 (Figure from Cohen)
  • 7. Seoul National University2018/1/27 - 7 - • How the negative values are treated in the literature Normal 50% fault 100% fault Staszewski, Wieslaw J., Keith Worden, and Geof R. Tomlinson. "Time–frequency analysis in gearbox fault detection using the Wigner–Ville distribution and pattern recognition." Mechanical systems and signal processing 11.5 (1997): 673-692. 327 cited “The negative values of the distribution were set to zero to avoid difficulties with the physical interpretation.” Baydar, Naim, and Andrew Ball. "A comparative study of acoustic and vibration signals in detection of gear failures using Wigner–Ville distribution." Mechanical systems and signal processing 15.6 (2001): 1091-1107. 272 cited Normal 25% fault 50% fault “To overcome this problem and reduce the presence of interference components, a smoothed version of the WVD (SPWVD) is used.” 8. The Wigner Distribution
  • 8. Seoul National University • Property (Refer to Cohen to check the proof) 4. Global average 2018/1/27 - 8 - • Definition (Related to the energy of the signals) ß Global average (due to marginal property) 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 < 𝑔[ 𝑡 + 𝑔2 𝜔 >= ' ' 𝑔[ 𝑡 + 𝑔2 𝜔 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡 = ∫ 𝑔[ 𝑡 𝑠(𝑡) 2 𝑑𝑡 + ∫ 𝑔2 𝜔 𝑆(𝜔) 2 𝑑𝜔 < 𝑔 𝑡, 𝜔 >= ' ' 𝑔 𝑡, 𝜔 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡
  • 9. Seoul National University • Property (Refer to Cohen to check the proof) 5. Local average • Instantaneous frequency and group delay can be derived from local averages of the Wigner distribution 2018/1/27 - 9 - • Definition (Related to the energy of the signals) ß Local average 𝜑 : phase 𝜓 : spectral phase Instantaneous frequency Group delay 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 < 𝜔 >:= 1 𝑠(𝑡) 2 ' 𝜔𝑊 𝑡, 𝜔 𝑑𝜔 < 𝑡 >Z= 1 𝑆(𝜔) 2 ' 𝑡𝑊 𝑡, 𝜔 𝑑𝑡 𝑡 ; < 𝜔 >:= 𝜑′(𝑡) ; < 𝑡 >Z= −𝜓′(𝜔)
  • 10. Seoul National University • Property (Refer to Cohen to check the proof) 6. Time and Frequency shift 2018/1/27 - 10 - • Definition (Related to the energy of the signals) 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 if 𝑠 𝑡 → 𝑒1Zn: 𝑠 𝑡 − 𝑡o then 𝑊 𝑡, 𝜔 → 𝑊(𝑡 − 𝑡o,𝜔 − 𝜔o) 𝑊st 𝑡, 𝜔 = 1 2𝜋 ' 𝑒01Zn :05/2 𝑠∗ (𝑡 − 𝑡o − 𝜏 2 ) ×𝑒1Zn :75/2 𝑠(𝑡 − 𝑡o + 5 2 )𝑒015Z 𝑑𝜏 = [ 23 ∫ 𝑠∗ (𝑡 − 𝑡o − 5 2 )𝑠(𝑡 − 𝑡o + 5 2 ) 𝑒015(Z0Zn) 𝑑𝜏 = 𝑊(𝑡 − 𝑡o, 𝜔 − 𝜔o)
  • 11. Seoul National University • Property (Refer to Cohen to check the proof) 7. Cross-term (Interference) • For multi-component signals, cross-terms come out due to quadratic calculation 2018/1/27 - 11 - • Definition (Related to the energy of the signals) Cross-terms 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝑠 𝑡 =𝑠1 𝑡 +𝑠2 𝑡 𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 𝑊12 𝑡, 𝜔 + 𝑊21 𝑡, 𝜔 where 𝑊12 𝑡, 𝜔 = ' 𝑠[ ∗ 𝑡 − 𝜏 2 𝑠2(𝑡 + 𝜏 2 )𝑒015Z 𝑑𝜏 𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 2Re {𝑊12 𝑡, 𝜔 } (Figure from Auger)
  • 12. Seoul National University2018/1/27 - 12 - • Definition (Related to the energy of the signals) 8. The Wigner Distribution • Property (Refer to Cohen to check the proof) 7. Cross-term (Interference) • For multi-component signals, cross-terms come out due to quadratic calculation 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 Cross-terms 𝑠 𝑡 =𝑠1 𝑡 +𝑠2 𝑡 𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 𝑊12 𝑡, 𝜔 + 𝑊21 𝑡, 𝜔 where 𝑊12 𝑡, 𝜔 = ' 𝑠[ ∗ 𝑡 − 𝜏 2 𝑠2(𝑡 + 𝜏 2 )𝑒015Z 𝑑𝜏 𝑊 𝑡, 𝜔 = 𝑊11 𝑡, 𝜔 + 𝑊22 𝑡, 𝜔 + 2Re {𝑊12 𝑡, 𝜔 } (Figure from Cohen)
  • 13. Seoul National University2018/1/27 - 13 - • Definition (Related to the energy of the signals) ü First let us make clear that it is not generally true that the cross terms produce undesirable effects. ~~~ In fact, since any signal can be broken up into a sum of parts in an arbitrary way, the cross terms can be neither bad nor good since they are not uniquely defined; they are different for different decompositions. The Wigner distribution does not know about cross terms, since the breaking up of a signal into parts is not unique. (P.126, Cohen) ü However, the localization and amplitude of these additional terms often make the use and interpretation of the representation difficult, or even impossible when the signal contains a large number of “elementary components”. Since these interference terms distribute the real part of the scalar product in the time-frequency plane, they distribute negative values when the scalar product is negative. (P. 148-149, Auger) 8. The Wigner Distribution • Property (Refer to Cohen to check the proof) 7. Cross-term (Interference) • Two difference views on cross-terms 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08
  • 14. Seoul National University • Property • Instantaneous frequency and group delay can be derived by local average. • The outputs could have negative values, which is counter-intuitive. • Suffers from the fact that confusing artifacts could be achieved for multicomponent signals (Cross-terms) 2018/1/27 - 14 - • Comparison between the Wigner distribution and the spectrogram Wigner distribution Spectrogram • Property • Instantaneous frequency and group delay can only be approximated. • The outputs always have positive values. • The multi-component could not be effectively resolved. (Window size) 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗(𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝐹 𝑥 𝑡, 𝜈; ℎ = ' 𝑥 𝑢 ℎ∗(𝑢 − 𝑡)𝑒01234z 𝑑𝑢 78 08
  • 15. Seoul National University2018/1/27 - 15 - • Smoothed-pseudo Wigner-Ville distribution (SPWVD): To solve cross-term problems WVD: PWVD: SPWVD: (Smoothing in frequency-domain) (Smoothing both in time- and frequency-domain) 8. The Wigner Distribution 𝑊 𝑥 𝑡, 𝜈 = ' 𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝑃𝑊 𝑥 𝑡, 𝜈 = ' ℎ(𝜏)𝑥 𝑡 + 𝜏 2 𝑥∗ (𝑡 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 𝑆𝑃𝑊 𝑥 𝑡, 𝜈 = ' ℎ(𝜏) ' 𝑔(𝑠 − 𝑡)𝑥 𝑠 + 𝜏 2 𝑥∗ (𝑠 − 𝜏 2 )𝑒012345 𝑑𝜏 78 08 78 08
  • 16. Seoul National University2018/1/27 - 16 - • Smoothed-pseudo Wigner-Ville distribution (SPWVD): To solve cross-term problems (figure from Auger) WVD PWVD SPWVD Smoothing in freq. Smoothing in time 8. The Wigner Distribution
  • 17. Seoul National University2018/1/27 - 17 - • Definition (Cohen) (Auger) Kernel function Parameterization function • Types of kernels • Product kernel: General case • Separable kernel 9. General Approach and the Kernel Method (The Cohen’s class) 𝐶 𝑡, 𝜔 = 1 4𝜋2 ' ' ' 𝑠∗ 𝑢 − 𝜏 2 𝑠 𝑢 + 𝜏 2 𝜙 𝜃, 𝜏 𝑒01}:015Z71}z 𝑑𝑢𝑑𝜏𝑑𝜃 𝐶~ 𝑡, 𝜐; 𝑓 = ' ' ' 𝑒123• s0: 𝑓(𝜉, 𝜏)𝑥 𝑠 + 𝜏 2 𝑥∗(𝑠 − 𝜏 2 )𝑒012345 𝑑𝜉𝑑𝑠𝑑𝜏 78 08 𝜙(𝜃, 𝜏) = 𝜙ƒ„ 𝜃𝜏 = 𝜙(𝜃𝜏) 𝜙 𝜃, 𝜏 = 𝜙[(𝜃)𝜙[(𝜏)
  • 18. Seoul National University2018/1/27 - 18 - • Some Distributions and Their Kernels (Table from Cohen) 9. General Approach and the Kernel Method (The Cohen’s class)
  • 19. Seoul National University2018/1/27 - 19 - • Basic properties related to the kernel • Marginals: Instantaneous Energy / Energy Density Spectrum Basic form Integrating wrt frequency For the integration to be instantaneous power ( )For frequency marginal For total energy 9. General Approach and the Kernel Method (The Cohen’s class) 𝐸 = ' ' 𝑊 𝑡, 𝜔 𝑑𝜔𝑑𝑡 = ' 𝑠(𝑡) 2 𝑑𝜏 𝑃 𝑡 = ' 𝑊 𝑡, 𝜔 𝑑𝜔 = 𝑠(𝑡) 2 𝐶 𝑡, 𝜔 = 1 4𝜋2 ' ' ' 𝑠∗ 𝑢 − 𝜏 2 𝑠 𝑢 + 𝜏 2 𝜙 𝜃, 𝜏 𝑒01}:015Z71}z 𝑑𝑢𝑑𝜏𝑑𝜃 '𝐶 𝑡, 𝜔 𝑑𝜔 = 1 2𝜋 ' ' ' 𝛿(𝜏)𝑠∗ 𝑢 − 𝜏 2 𝑠 𝑢 + 𝜏 2 𝜙 𝜃, 𝜏 𝑒1}(z0:) 𝑑𝑢𝑑𝜏𝑑𝜃 = 1 2𝜋 ' '𝜙 𝜃, 0 𝑠(𝑢) 2 𝑒1}(z0:) 𝑑𝜃𝑑𝑢 1 2𝜋 '𝜙 𝜃, 0 𝑒1}(z0:) 𝑑𝜃 = 𝛿(𝑡 − 𝑢) 𝜙 𝜃, 0 =1 𝜙 0, 𝜏 =1 𝜙 0,0 =1
  • 20. Seoul National University2018/1/27 - 20 - • Basic properties related to the kernel • Time and frequency shift • Scaling invariance • Local average • Global average • … 9. General Approach and the Kernel Method (The Cohen’s class) 𝐶st 𝑡, 𝜔 = 1 4𝜋2 ' ' ' 𝑒01Zn(z0 5 2 0:n) 𝑒1Zn(z7 5 2 0:n) × 𝑠∗ 𝑢 − 5 2 − 𝑡o 𝑠 𝑢 + 5 2 − 𝑡o 𝜙 𝜃, 𝜏 𝑒01}:015Z71}z 𝑑𝑢𝑑𝜏𝑑𝜃 = 1 4𝜋2 ' ' ' 𝜙 𝜃, 𝜏 𝑠∗ 𝑢 − 𝜏 2 𝑠 𝑢 + 𝜏 2 𝑒01}:015(Z0Zn)71}(z7:n) 𝑑𝑢𝑑𝜏𝑑𝜃 = 1 4𝜋2 ' ' ' 𝜙 𝜃, 𝜏 𝑠∗ 𝑢 − 𝜏 2 𝑠 𝑢 + 𝜏 2 𝑒01}(:0:n)015(Z0Zn)71}z 𝑑𝑢𝑑𝜏𝑑𝜃 = 𝐶 𝑡 − 𝑡o, 𝜔 − 𝜔o
  • 21. Seoul National University2018/1/27 - 21 - • Objective: To maintain the good properties of the Wigner distribution 11. Kernel Design for Reduced Interference where *Weak finite support *Strong finite support For product kernel, 𝜙(𝜃, 𝜏) = 𝜙ƒ„ 𝜃𝜏 = 𝜙(𝜃𝜏) (Table from Cohen) ℎ 𝑡 = 1 2𝜋 '𝜙 𝑥 𝑒1~: 𝑑𝑥 ; 𝜙 𝜃𝜏 = 'ℎ 𝑡 𝑒01}5: 𝑑𝑡 𝑃 𝑡, 𝜔 = 0 for 𝑡 outside 𝑡[, 𝑡2 if 𝑠 𝑡 is zero outside 𝑡[, 𝑡2 𝑃 𝑡, 𝜔 = 0 for 𝜔 outside 𝜔[, 𝜔2 if 𝑆 𝜔 is zero outside 𝜔[, 𝜔2 𝑃 𝑡, 𝜔 = 0 if 𝑠 𝑡 = 0 for a particular time 𝑃 𝑡, 𝜔 = 0 if 𝑆 𝜔 = 0 for a particular frequency
  • 22. Seoul National University2018/1/27 - 22 - • Choi-Williams method • Properties • Product kernel • Both marginal are satisfied (The energy spectral density 𝑺(𝝎) 𝟐 and the instantaneous power 𝒔(𝒕) 𝟐 can be obtained) • Distribution 12. Some distributions *H.I. Choi: Faculty of the Global School Of Media at the Soongsil University *W.J. Williams: Faculty of the Department of Electrical Engineering and Computer Science at the University of Michigan (For frequency marginal) (For time marginal) Kernel function ! ", $ = 1 4() * ** +∗ - − / 2 + - + / 2 2 3, / 45678569:;67< =-=/=3 𝜙 𝜃, 𝜏 = 𝑒0}‘5‘/’ 𝜙 0, 𝜏 = 1 𝜙 𝜃, 0 = 1 𝑃“” 𝑡, 𝜔 = 1 4𝜋J/2 ' ' 1 𝜏2 /𝜎 exp (𝑢 − 𝑡)2 4𝜏2/𝜎 − 𝑗𝜏𝜔 × 𝑠∗ 𝑢 − 5 2 𝑠 𝑢 + 5 2 𝑑𝑢𝑑𝜏
  • 23. Seoul National University2018/1/27 - 23 - • Choi-Williams method: Examples • For the sum of two sine waves ( ), the distribution will be calculated as where à The distribution would have a large peak at 𝝎 = 𝝎 𝟏 7𝝎 𝟐 𝟐 for large 𝝈 12. Some distributions Wigner distribution C-W with a large 𝝈 C-W with a small 𝝈 *C-W becomes WD for 𝜎 → ∞ 𝜙 𝜃, 𝜏 = 𝑒0}‘5‘/’ 𝑠 𝑡 = 𝐴[ 𝑒1Z˜: + 𝐴2 𝑒1Z‘: 𝐶“” 𝑡, 𝜔 = 𝐴[ 2 𝛿 𝜔 − 𝜔[ + 𝐴2 2 𝛿 𝜔 − 𝜔2 + 2𝐴[ 𝐴2 cos[ 𝜔2 − 𝜔[ 𝑡]𝜂(𝜔, 𝜔[, 𝜔2, 𝜎) 𝜂 𝜔, 𝜔[, 𝜔2, 𝜎 = 1 4𝜋 𝜔[ − 𝜔2 2/𝜎 exp 𝜔 − 1 2 𝜔[ + 𝜔2 2 4𝜋 𝜔[ − 𝜔2 2/𝜎 Figure from Cohen
  • 24. Seoul National University2018/1/27 - 24 - • Choi-Williams method 12. Some distributions WD C-W Spectrogram Figure from Cohen
  • 25. Seoul National University2018/1/27 - 25 - • Born-Jordan Distribution: Reduced interference • Zhao-Atlas-Marks Distribution: Reduced interference by placing cross-terms under the self-terms 12. Some distributions 𝜙 𝜃, 𝜏 = sin(𝑎𝜃𝜏) 𝑎𝜃𝜏 𝜙š›œ 𝜃, 𝜏 = 𝑔 𝜏 𝜏 sin(𝑎𝜃𝜏) 𝑎𝜃𝜏 Figure from Cohen
  • 26. Seoul National University2018/1/27 - 26 - Literature review Feng, Zhipeng, Ming Liang, and Fulei Chu. "Recent advances in time–frequency analysis methods for machinery fault diagnosis: A review with application examples." Mechanical Systems and Signal Processing 38.1 (2013): 165-205. 283 cited • Linear time–frequency representation STFT WT Signal: 𝑥 𝑡 = sin 2𝜋𝑓 ¡¢£ 𝑡 + 2 cos 2𝜋𝑓¤¥¦¦¡£¦ 𝑡 + 153.6 cos 2𝜋𝑓«¬ 𝑡 + 𝑛(𝑡)
  • 27. Seoul National University2018/1/27 - 27 - Literature review Feng, Zhipeng, Ming Liang, and Fulei Chu. "Recent advances in time–frequency analysis methods for machinery fault diagnosis: A review with application examples." Mechanical Systems and Signal Processing 38.1 (2013): 165-205. 283 cited • Bilinear time–frequency distribution WVD SPWVD C-H Signal: 𝑥 𝑡 = sin 2𝜋𝑓 ¡¢£ 𝑡 + 2 cos 2𝜋𝑓¤¥¦¦¡£¦ 𝑡 + 153.6 cos 2𝜋𝑓«¬ 𝑡 + 𝑛(𝑡)
  • 28. Seoul National University2018/1/27 - 28 - Literature review • Basic principles of gear fault diagnosis à Based on side-band detection *Feng, Zhipeng, and Ming Liang. "Fault diagnosis of wind turbine planetary gearbox under nonstationary conditions via adaptive optimal kernel time–frequency analysis." Renewable Energy 66 (2014): 468-477. 56 cited * * The interference terms from WVD would make it difficult to diagnose the fault in the system