You are on page 1of 18

A Data-driven Approach to Monitoring Wind

Turbine Faults
Graduate Student: Zijun Zhang and Anoop Verma
PI: Prof. Andrew Kusiak
Intelligent Systems Laboratory
The University of Iowa
September 21st, 2011
The University of Iowa

Intelligent Systems Laboratory

Outline
Gearbox fault identification
1. Analysis of Case 2a
2. Analysis of Case 2b
3. Analysis of Case 2c

Monitoring gearbox fault


1. Data-driven jerk prediction model
2. Monitoring models

Discussion
The University of Iowa

Intelligent Systems Laboratory

Gearbox Fault Identification


Jerk the rate of acceleration change, and it is often used
to indicate the excitement of vibration.
at at T
J
T

da
at at T
a at at T

= lim

=
J=
, a = at at T and t = t (t T )

0
t t (t T )
dt
T

Data processing Convert 40kHz data to 1/15kHz. In the


data conversion, estimate max, min and standard deviation.

The University of Iowa

Intelligent Systems Laboratory

Gearbox Fault Identification


Analysis of Case 2a
Time domain analysis

Maximum of change of generator speed


Pattern of maximal Jerk for AN4

600
500

25

400

20

300

Maximal of jerk

Maximal rate of speed (rpm/15-s)

200
100
0
1

9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Time (15 s interva ls)
Speed of encoder on HSS

15
10
5
0
1

9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Time (15 s intervals)

The University of Iowa

Intelligent Systems Laboratory

Gearbox Fault Identification


Analysis of Case 2a
Frequency domain analysis Fourier transformation with time window

0.035

0.04

0.025

0.035

0.02

0.03
0.025

0.015
Power

Power

0.03

5th minute

0.01

0.02
0.015

0.005
0.01

500

1000

1500
2000
2500
Frequency (Hz)

2nd minute

The University of Iowa

3000

3500

4000

0.005
0

500

1000

1500
2000
2500
Frequency (Hz)

3000

3500

Intelligent Systems Laboratory

4000

Gearbox Fault Identification


Analysis of Case 2b
Time domain analysis
Comparison between 2a and 2b
Sensors

AN1

AN2

AN3

AN4

AN5

AN6

AN7

AN8

AN9

AN10

AN11

AN12

Mean of maximum jerk of 2b

0.6

0.6

4.4

5.0

7.2

13.0

9.6

17.4

18.9

6.1

9.9

23.4

Mean of maximum jerk of 2a

0.7

0.8

9.4

14.2

3.1

5.8

4.6

8.3

8.4

5.3

6.7

10.4

Correlation Coefficient Analysis

Clustering Analysis

Sensors

AN1

AN2

AN3

AN4

AN5

AN6

AN7

AN8

AN9

AN10

AN11

AN12

Sensors

Index of clusters

AN1

1.00

0.98

0.53

0.20

0.02

0.14

0.00

-0.11

0.16

0.16

0.16

0.03

AN1

AN2

0.98

1.00

0.59

0.06

0.18

0.30

0.17

0.05

0.32

0.25

0.33

0.19

AN2

AN3

0.53

0.59

1.00

0.09

0.31

0.53

0.41

0.19

0.62

0.77

0.48

0.32

AN3

AN4

0.20

0.06

0.09

1.00

-0.88

-0.78

-0.85

-0.94

-0.70

-0.11

-0.76

-0.89

AN4

AN5

0.02

0.18

0.31

-0.88

1.00

0.94

0.97

0.97

0.90

0.50

0.92

0.97

AN5

AN6

0.14

0.30

0.53

-0.78

0.94

1.00

0.98

0.93

0.99

0.60

0.95

0.96

AN6

AN7

0.00

0.17

0.41

-0.85

0.97

0.98

1.00

0.97

0.96

0.56

0.94

0.97

AN7

AN8

-0.11

0.05

0.19

-0.94

0.97

0.93

0.97

1.00

0.88

0.39

0.90

0.98

AN8

AN9

0.16

0.32

0.62

-0.70

0.90

0.99

0.96

0.88

1.00

0.67

0.92

0.93

AN9

AN10

0.16

0.25

0.77

-0.11

0.50

0.60

0.56

0.39

0.67

1.00

0.59

0.42

AN10

AN11

0.16

0.33

0.48

-0.76

0.92

0.95

0.94

0.90

0.92

0.59

1.00

0.92

AN11

AN12

0.03

0.19

0.32

-0.89

0.97

0.96

0.97

0.98

0.93

0.42

0.92

1.00

AN12

The University of Iowa

Intelligent Systems Laboratory

Gearbox Fault Identification


Analysis of Case 2b
Time domain analysis
Cluster 1
0.00
2.31
3.41

Cluster 3
3.41
5.72
0.00

1.24

1.64
1.62
1.6
1.58
1.56
1.54
1.52
1.5
1.48
1.46
1.44
1.42

Standard devuiation of jerk

Mean of jerk

Cluster 1
Cluster 2
Cluster 3

Cluster 2
2.31
0.00
5.72

1.22
1.2
1.18
1.16
1.14
1.12
1.1
1.08

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Time (15 s intervals)
AN7

The University of Iowa

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Time (15 s intervals)
AN7

Intelligent Systems Laboratory

Gearbox Fault Identification


Frequency domain analysis

0.16

0.12

0.14

0.1

0.12
0.1

Power

Power

0.08

0.06

0.08
0.06

0.04

0.04
0.02

0.02

500

1000

1500
2000
2500
Frequency (Hz)

The University of Iowa

3000

3500

4000

500

1000

2500
2000
1500
Frequency (Hz)

3000

3500

4000

Intelligent Systems Laboratory

Gearbox Fault Identification


Analysis of Case 2c
Time domain analysis

Comparison between 2a, 2b and 2c


Sensor

AN1

AN2

AN3

AN4

AN5

AN6

AN7

AN8

AN9

AN10

AN11

AN12

Mean of maximum jerk in 2b

0.6

0.6

4.4

5.0

7.2

13.0

9.6

17.4

18.9

6.1

9.9

23.4

Mean of maximum jerk in 2a

0.7

0.8

9.4

14.2

3.1

5.8

4.6

8.3

8.4

5.3

6.7

10.4

Mean of maximum jerk in 2c

3.22

5.11

6.09

8.73

8.53

15.18

12.80

20.71

22.57

7.50

10.53

28.33

Correlation Coefficient Analysis

Clustering Analysis

Sensor

AN1

AN2

AN3

AN4

AN5

AN6

AN7

AN8

AN9

AN10

AN11

AN12

AN1

1.00

0.83

0.32

-0.73

0.50

0.56

0.49

0.67

0.49

0.48

0.64

0.56

AN1

AN2

0.83

1.00

0.76

-0.91

0.88

0.90

0.87

0.92

0.87

0.86

0.91

0.91

AN2

AN3

0.32

0.76

1.00

-0.67

0.94

0.92

0.95

0.78

0.96

0.95

0.80

0.90

AN3

AN4

-0.73

-0.91

-0.67

1.00

-0.85

-0.90

-0.84

-0.97

-0.83

-0.80

-0.94

-0.91

AN4

AN5

0.50

0.88

0.94

-0.85

1.00

0.99

0.99

0.92

0.99

0.98

0.92

0.98

AN5

AN6

0.56

0.90

0.92

-0.90

0.99

1.00

0.99

0.95

0.98

0.97

0.95

0.99

AN6

AN7

0.49

0.87

0.95

-0.84

0.99

0.99

1.00

0.91

0.99

0.98

0.92

0.98

AN7

AN8

0.67

0.92

0.78

-0.97

0.92

0.95

0.91

1.00

0.91

0.88

0.96

0.96

AN8

AN9

0.49

0.87

0.96

-0.83

0.99

0.98

0.99

0.91

1.00

0.98

0.90

0.98

AN9

AN10

0.48

0.86

0.95

-0.80

0.98

0.97

0.98

0.88

0.98

1.00

0.88

0.96

AN10

AN11

0.64

0.91

0.80

-0.94

0.92

0.95

0.92

0.96

0.90

0.88

1.00

0.96

AN11

AN12

0.56

0.91

0.90

-0.91

0.98

0.99

0.98

0.96

0.98

0.96

0.96

1.00

AN12

The University of Iowa

Sensors

Index of clusters

Intelligent Systems Laboratory

Gearbox Fault Identification


Analysis of Case 2c
Time domain analysis
Cluster 1
Cluster 2
Cluster 3

Cluster 1
0.00
2.26
3.26

Cluster 2
2.26
0.00
5.52

1.65

2.2

Standard deviation of jerk

2.15
2.1

Mean of jerk

Cluster 3
3.26
5.52
0.00

2.05
2
1.95
1.9
1.85
1.8

1.6
1.55
1.5
1.45
1.4
1.35

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Time (15 s intervals)
AN7

The University of Iowa

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Time (15 s intervals)
AN7

Intelligent Systems Laboratory

Monitoring Gearbox Fault


Data driven single-signal jerk prediction model
J AN7,t = f ( J AN7,t T , J AN7,t 2T , J AN7,t 3T , J AN7,t 9T , J AN7,t 10T )
Parameter description
Parameter
JAN7,t T
JAN7,t 2T
JAN7,t 3T
JAN7,t 9T
JAN7,t 10T

The University of Iowa

Importance
100
96
93
93
92

Parameter
JAN7,t 8T
JAN7,t 4T
JAN7,t 5T
JAN7,t 7T
JAN7,t 6T

Importance
92
90
90
89
82

Intelligent Systems Laboratory

Monitoring Gearbox Fault


Test results
Algorithm
NN
NNE
BT
RF
CART
SVM
KNN

MAE
0.0445
0.0446
0.0427
0.0505
0.0573
0.0408
0.0505

SDofAE
0.0329
0.0331
0.0315
0.0361
0.0391
0.0296
0.0388

MAPE
0.0207
0.0208
0.0200
0.0235
0.0266
0.0192
0.0235

SDofAPE
0.0150
0.0151
0.0144
0.0163
0.0176
0.0138
0.0178

2.25

First 100 test points

Jerk (m/s3)

2.2
2.15
2.1
2.05
2
1.95

1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97

1.9
Time (0.1-s interva ls)
Observed va lue

The University of Iowa

Predicted va lue

Intelligent Systems Laboratory

Monitoring Gearbox Fault


Modified x-bar chart
m

=
t [ ( yi yi )] / m
i =1

=
st

[ (( y i yi ) t ) 2 ] /(m 1)
i =1

UCL=
t + c( st / q )
x
CL x = t
=
LCL x max{0, t c( st / q )}
The UCLx, CLx and LCLx represent the upper control limit, centerline, and lower control limit
of the modified x-bar chart, respectively. The parameter c controls the width between the upper
and lower control limit. The parameter q denotes the number of sampling data points. The
parameter t represents the mean of the test dataset, st describes the standard deviation of the
test dataset and m is the total number of data points in the test dataset

The University of Iowa

Intelligent Systems Laboratory

Monitoring Gearbox Fault


Modified x-bar chart

Test of process shift


0.14

0.07

0.12

Absolute percenta ge error

0.08
0.06
0.05
0.04
0.03
0.02
0.01

0.1
0.08
0.06
0.04
0.02
1
21
41
61
81
101
121
141
161
181
201
221
241
261
281
301
321
341
361
381
401
421
441
461
481

0
1
21
41
61
81
101
121
141
161
181
201
221
241
261
281
301
321
341
361
381
401
421
441
461
481

Absolute percenta ge error

Training control chart

Time (0.1s intervals)

Time (0.1s intervals)


Residuals

Centerline

Upper bound

The University of Iowa

Lower bound

Residuals

Centerline

Upper bound

Lower bound

Intelligent Systems Laboratory

Monitoring Gearbox Fault


EWMA (Exponentially Weighted Moving Average) chart
zi = xi + (1 ) zi 1
UCL EWMA = t + Lst

[1 (1 ) 2i ]
2

CL EWMA = t
LCL
max{0, t Lst
=
EWMA

[1 (1 ) 2i ]}
2

Parameter z presents the EWMA, x is the prediction residual and i is the index of a
data point. L and are two parameters that need to be arbitrarily set to develop the
EWMA chart.

The University of Iowa

Intelligent Systems Laboratory

Monitoring Gearbox Fault


Results of EWMA chart

Absolute percentage error

0.06
0.05
0.04
0.03
0.02
0.01
0
1

11 13 15 17 19 21 23 25 27 29
Time (0.1-s intervals)

EWMA

The University of Iowa

Centerline

Upper bound

Lower bound

Intelligent Systems Laboratory

Discussion
Data collected from the nacelle fixed on the
floor. The data might not fully reflect the
vibration of the gearbox in the real wind
turbine.
In the industry, wind turbines usually
contain only two accelerometers, some
might have six accelerometers.

The University of Iowa

Intelligent Systems Laboratory

Sensors
Accelerometer No. 1
Accelerometer No. 2
Accelerometer No. 3
Accelerometer No. 4
Accelerometer No. 5

Symbols
AN1
AN2
AN3
AN4
AN5

Accelerometer No. 6

AN6

Accelerometer No. 7

AN7

Accelerometer No. 8

AN8

Accelerometer No. 9

AN9

Accelerometer No. 10
Accelerometer No. 11
Accelerometer No. 12

AN10
AN11
AN12

Generator speed

Low speed shaft torque

The University of Iowa

Locations
Main bearing radial
Main bearing axial
Ring gear radial 6 oclock
Ring gear radial 12 oclock
Low speed shaft radial
Intermediate speed shaft
radial
High speed shaft radial
High speed shaft upwind
bearing radial
High speed shaft
downwind bearing radial
Carrier downwind radial
Generator upwind radial
Generator downwind axial
Encoder on high speed
shaft
Strain gauges on low speed
shaft

Plus acceleration directions


+Z and -Y
-X
-Z
+Z
+Y and -Z

Units
m/s2
m/s2
m/s2
m/s2
m/s2

+Y and -Z

m/s2

+Y and +Z

m/s2

+Z

m/s2

+Z

m/s2

+Y
+Z and -Y
+Z and -Y

m/s2
m/s2
m/s2

Null

rpm

Null

kNm

Intelligent Systems Laboratory

You might also like