14. - LD: “Logical Device”
Correspón a un dispositiu (Aerogenerador, Torre
meteorológica,…)
- LN: “Logical Node”
Correspón a una parte del dispositivo (Torre, Góndola,
Convertidor,…)
Conceptes
20. Construcció nom variables
LN.[NomVar].[TipoVar]
LN: Nombre del nodo lógico.
[NomVar]: Nombre de variable. Puede contener uno o varios niveles,
cada uno de ellos separados por puntos (.), dependiendo de la
variable.
[TipoVar]: Tipo de la variable. Puede contener uno o varios niveles,
cada uno de ellos separados por puntos (.), dependiendo del tipo de la
variable.
Ex: WGDC.TrfGri.PhV.phsA.cVal.avgVal.f
WGDC: Nodo Grid
TrfGri.PhV.phsA: Grid Trifásico.Voltaje de fase. Fase A
cVal.avgVal.f: Valor de variable. Valor medio. Float
21. Algunas variables Furhländer
WNAC.Wdir1.avgVal.f
WNAC: Nodo Nacelle
Wdir1: Wind direction 1
avgVal.f: Valor medio. Float
WNAC.Wdir1.minVal.f
WNAC: Nodo Nacelle
Wdir1: Wind direction 1
minVal.f: Valor mínimo. Float
WTUR.ExtPwrReactSp.maxVal.f
WTUR: Nodo Wind Turbine
ExtPwrReactSp: External Power Reactive Speed
maxVal.f: Valor máximo. Float
WTUR.ExtPwrReactSp.maxVal.f
WTUR: Nodo Wind Turbine
ExtPwrReactSp: External Power Reactive Speed
maxVal.f: Valor máximo. Float
28. Technologia
Physical Variables:
Vibration Analysis
Motor Current Signature Analysis
Voltage Measurements
Acoustic Emission Measurements
Temperature Monitoring
Signal Processing Techniques:
Frequency Analysis
Time analysis
Time-Frequency Analysis
Decision Support Systems:
Neural Networks
Fuzzy
SVM
Random Forest Algorithms
Current ConditionMonitoringTechniques
35. Datasets that are linearly separable with some noise work out
great:
But what are we going to do if the dataset is just too hard?
How about… mapping data to a higher-dimensional space:
0 x
0 x
0 x
x2
Classificadors No-Lineals, SVM
36. General idea: the original input space can always be mapped
to some higher-dimensional feature space where the training
set is separable:
Φ: x → φ(x)
Classificadors No-Lineals, SVM
39. • Broken Bearings. (R.R Schoen
and Others 1994)
where nb number of balls, fi,0 fault
vibration frequencies, fr rotating
frequency Hz, bd ball diameter, pd
Race diameter, & β ball angle.
Relevant Frequencies Bearing
o,isbng mfff
cos
pd
bd
f
n
f rb
o,i 1
2
40. Relevant Frequencies Generator
Fault frequencies analyzed on the Gearbox can be complemented by
measurements on the generator.
Generator fault frequencies shall be analyzed using vibration or current (Motor
Current Signature Analysis) measurements.
Fault condition on the gearbox usually appears as an eccentricity fault on the
generator, this fault is usually one of the most relevant indicators to address
fault condition analysis.
Further results are related to MCSA condition monitoring results.
41. Relevant Frequencies Generator
• Eccentricity fault (Thomson 1988)
where m=1,2,3,… harmonic number, p is the pair of poles, s the slip, y fs electric
frequency.
p
s
mff secc
1
1
42. Relevant Frequencies Generator
• Broken rotor bars, just for induction (Kliman 1988, Benbouzid 1995)
where l/p= 1,5,7,11,13,… are harmonic motor characteristics
s
p
s
lff sbrb
1
0 25 50 75 100 125 150 175 200 225 250
0
0.05
0.1
0.15
0.2
2.8 A
Magnitude(A)
Frequency (Hz)
43. Relevant Frequencies Generator
• Shortcircuits (Thomson 1988, 1995)
– Low frequencies
k=0,1,3,5,...
ks
p
m
ff sstl 1
0 50 100 150 200 250 300 350
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
Frequency (Hz)
Magnitude(A)
Ia
Ib
Ic
2.76 A
2.38 A
2.62 A
44. Relevant Frequencies Generator
400 500 600 700 800 900 1000 1100 1200
0
0.002
0.004
0.006
0.008
0.01
Frequency (Hz)
Magnitude(A)
Ia
Ib
Ic
p
s
mZff ssth
1
1 2
• Short Circuits (Rosero - Cusidó 2006)
– Medium frequencies
where Z2 is the number of rotor slots & k=0,1,3,5,...
45. Time- frequency Transformation
Applied to transient analysis improving the resolution and accuracy for the
fault detection
• Short Time Fourier Transform: is
the time dependant fourier
transform
• It applies a temporal window in
wich the FT is performed
dtttfbfG bwdwwdw ,
:,
tj
wdw ebttfbfG wdw
,,
wdw
2
wdw2
wdw
2
2b
0b
2
wdw2
wdw
2
wdw2
1b t
0
1
47. Time- frequency Transformation
• Wavelet Transform: Wavelet transform
decomposes the signal as a sum of
different wavelet signals shifted and
scaled. Those signals are know as
“mother” wavelet.
• The Decomosition algorithm
decomponds the signal in a diadic way
• The output of the transformation is the
time evolution of each decomposition
or detail.
ndnanx
J
jj
kj
k
kj
k
kjkj
1
,,,,
0
00
g[n]
h[n]
2
x[n]
2
g[n]
h[n]
2
2
g[n]
h[n]
2
2
Level 1 detail coefficients
Scale 2J-1
Level 2 detail coefficients
Scale 2J-2
Level 3 detail coefficients
Scale 2J-3
Level 1 detail coefficients
Scale 2J-3
g[n]
h[n]
222
x[n]
222
g[n]
h[n]
222
222
g[n]
h[n]
222
222
Level 1 detail coefficients
Scale 2J-1
Level 2 detail coefficients
Scale 2J-2
Level 3 detail coefficients
Scale 2J-3
Level 1 detail coefficients
Scale 2J-3
Approx.
Level 3
Detail
Level 3
Detail
Level 2
Detail
Level 1
fs/2fs/4 ffs/8fs/160
Approx.
Level 3
Detail
Level 3
Detail
Level 2
Detail
Level 1
fs/2fs/4 ffs/8fs/160
48. Feature /
Analysis
Method
Fourier Fast
Transform
(FFT)
Wigner Ville
distribution
(WVD)
Continuous
Wavelet
Analysis
(CWT)
Discrete
Wavelet
Analysis
(DWT)
High order
spectral
Analysis
(HOSA)
Hilbert–Huang
Transform
(HHT)
Basis A priori A priori A priori A priori Adaptive
Frequency
Convolution
global
Convolution
global
Convolution global
Differentiation
:
Local
Presentation
Energy -
Frequency
Energy -
Frequency
Energy -
time -
frequency
Energy -
time -
frequency
Energy - time –
frequency
Energy - time
- frequency
No-linear No No No No Yes Yes
No-
Stationary
No Yes Yes Yes Yes Yes
Feature
Extraction
No No Yes No Yes
Advantage or
Disadvantag
e
Fixed
window
width, fixed
method a-
priori and
lineal
analysis.
Suffers
from cross
term
interference
and
Aliasing
problem.
Size of the basic
wavelet function is
constant.
Uniform resolution
and non-adaptive
nature.
It is adaptive
and highly
efficient.
Tool for
nonlinear and
non-stationary
analysis.
Signal Processing Techniques
51. SmartOpex
GMAO and Operations Platform
SmartOpex is the low cost implementation for
monitoring operations at a wind farm. It is
completely customizable to the clients, which
has an ERP or not, providing solutions to the
needs of management.
It consists in two parts, one with internet access
via a computer follows all the activities in the site
and allows the assignment of these tasks to the
maintenance teams. Using this application and
monitoring the operations, Smart Opex gives
typical indicators such as MTTR, MTBF, failure
rates, delays in preventive, downtime and lost
track of the hours of work. Smart Opex also
gives the working hours in the site.