45
1 Lunch lecture series of Monitoring Community 10 th May 2019 Automatic detection of rail surface defects using video image: A case study in the Dutch railways Dr. Alfredo Núñez Vicencio Assistant Professor, TU Delft Section of Railway Engineering

Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

1

Lunch lecture series of Monitoring Community10th May 2019

Automatic detection of rail surface defects using

video image: A case study in the Dutch railways

Dr. Alfredo Núñez VicencioAssistant Professor, TU Delft

Section of Railway Engineering

Page 2: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

2

1) Video Image based Monitoring of Rail

2) Deep Neural Networks for Rail Monitoring

3) Making use of the data: Risk analysis

4) Conclusions

Outline

Page 3: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

3

1) Video Image based Monitoring of Rail

2) Deep Neural Networks for Rail Monitoring

3) Making use of the data: Risk analysis

4) Conclusions

Outline

Page 4: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

4

Railway Infrastructure

Page 5: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

5

Railway Infrastructure

Risk

RiskRisk

Risk

Risk

Risk

Risk

Risk

Risk

Risk

Risk

Risk

Page 6: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

6

Defects in rails

Light Moderate Severe Two squats at a thermite weld

Initiating squat

Severe

Squats

CorrugationInsulated joint with

plastic surface degradation

Wheel burns

Damaged welds

…..

Page 7: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

7

In The Netherlands (about 7000 km of tracks)

Almost no time for monitoring and maintenance

Page 8: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

8

Video

Page 9: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

9

What do we need?

Page 10: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

10

What do we need?A method that can tell us whether the image is:

Healthy rail

Insulated rail joint

Rail surface defect

Page 11: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

11

What do we need?A method that can tell us whether the image is:

Healthy rail

Insulated rail joint

Rail surface defect

Page 12: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

12

1) Video Image based Monitoring of Rail

2) Deep Neural Networks for Rail Monitoring

3) Making use of the data: Risk analysis

4) Conclusions

Outline

Page 13: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

13

Images: dimensions

How many possible images do we have?

Grey scale (265 levels)

Dimensions266 x 224

Dimensions12 x 10

Answer: (12 x 10) 265

Grey scale

About 1.8 x 10532

possible images

Page 14: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

14

Images: dimensions

How many possible images do we have?

Grey scale (265 levels)

Dimensions266 x 224

Dimensions12 x 10

Answer: (12 x 10) 265

Grey scale

About 1.8 x 10532

possible images

The estimated number of atoms in the observable universe (1080)

M87's black hole is 53.5 million light-years away

(5.014 × 1026 millimeters)

But don’t worry, the Interesting features of the

image lies in a lower dimension

Page 15: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

15

Features hierarchy

Image Contour Edge Shapes

Smaller Dimensions

Page 16: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

16

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

Classification methods

Page 17: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

17

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

Classification methods

Page 18: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

18

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

0 0x 0 0x

Decision boundary

Classification methods

Page 19: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

19

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

0 0x 0 0x

Decision boundary

Classification methods

?

Page 20: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

20

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

0 0x 0 0x

Decision boundary

Classification methods

-

Page 21: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

21

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

0 0x 0 0x

Decision boundary

Classification methods

-

Page 22: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

22

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

0 0x 0 0x

Decision boundary Fe

atur

e 1

Feature 2

-

--

-

-

-

+

+

+

+

+

++

+ +

+

--

-

-

Classification methods

-

Page 23: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

23

Feat

ure

1

Feature 2

-

--

-

-

-

-

+

+

+

+

+

++

+

0 0x 0 0x

Decision boundary Fe

atur

e 1

Feature 2

-

--

-

-

-

+

+

+

+

+

++

+ +

+

--

-

-

Decision boundary

, 0f x , 0f x

Classification methods

-

Page 24: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

24

Possible methods

Source: http://yann.lecun.com/exdb/mnist/2015

0

2

4

6

8

10

12

14

LinearClassifiers

K-NearestNeighbors

BoostedStumps

SVMs Neural nets Convolutionalnets

test

err

or (

%)

MNIST 28x28 images, 60,000 train, 10,000 test

Page 25: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

25

Deep learning

Classic

Input

Output

Mapping (Hand designed)

ClassicMachine Learning

Input

Output

Mapping (Hand designed)

Deep Learning

Input

Output

Learned mapping

Tim de Bruin and Robert Babuska, “Artificial Neural Networks 1”. Course: Knowledge-Based Control Systems (SC42050), TUDelft.

Learned mapping

Learned features

Learned features

……

Page 26: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

26

Deep learning: learned features

https://ai.googleblog.com/2017/11/feature-visualization.html

Page 27: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

27

Deep learning: learned mapping

R. Babuska, Knowledge based Control Systems, Lecture notes.

McCulloch and Pitts modelled the behavior of a single neuron in

1943. They called this mathematical model a

Perceptron.

Page 28: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

28

1x

2x

nx

1w

2w

nw

z z y...

.

.

.

Artificial neuron

Deep learning: learned mapping

Page 29: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

29

1x

2x

nx

1w

2w

nw

z z y...

.

.

.

Inputs of the neuron

Adaptive weights (synaptic strength)

Activation function

For Perceptron

z

Output of the neuron

Deep learning: learned mapping

Page 30: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

30

Deep learning: learned mapping

R. Babuska, Knowledge based Control Systems, Lecture notes.

Paul Werbos, the father of backpropagation, since 1974 we

can train neural networks

Page 31: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

31S. Faghih-Roohi, S. Hajizadeh, A. Núñez, R. Babuska, and B. De Schutter, “A deep learning approach for detection of rail defects”. Proceedings of the IEEE World Congress on Computational Intelligence, IEEE WCCI 2016, 2016 International Joint Conference on

Neural Networks (IJCNN), Vancouver, Canada, 25-29 July, 2016, pp. 2584-2589.

Detection of rail defects

Page 32: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

32

Image data• The dataset consists of 4220 samples, of which 3170 are

normal, and roughly 1000 are defects.

• We train a convolutional neural network model with 80% of the data, and test with the remaining 20% (in 5 folds). Here is the averaged result of the test:

Accuracy = 0.9870

Predicted normal Predicted defect

Normal samples 635 1

Defects 10 197

32

Page 33: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

33

False detections (image data)

Defects not detected

Images from INSPECTATION

Page 34: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

34

Hits 1 (image data)True Positive

TU DELFT

Images from INSPECTATION

Page 35: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

35

Hits 2 (image data)

TU DELFT

Images from INSPECTATION

True Positive

Page 36: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

36

Classification of types• We also tried to classify the defects into 2 categories of

spots/light vs. medium/severe.

TU DELFT

Page 37: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

37

1) Video Image based Monitoring of Rail

2) Deep Neural Networks for Rail Monitoring

3) Making use of the data: Risk analysis

4) Conclusions

Outline

Page 38: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

38A. Jamshidi, S. Faghih-Roohi, S. Hajizadeh, A. Núñez, R. Babuška, R. Dollevoet, Z. Li and B. De Schutter, “A big data analysis

approach for rail failure risk assessment”. Risk Analysis, Volume 37, Issue 8, August 2017, Pages: 1495-1507.

A big data analysis approach is used to automatically detect squats from rail

images.

A Bayesian model is employed to estimate the failure probability.

Page 39: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

39

Architecture of the proposed DCNN model

A. Jamshidi, S. Faghih-Roohi, S. Hajizadeh, A. Núñez, R. Babuška, R. Dollevoet, Z. Li and B. De Schutter, “A big data analysis approach for rail failure risk assessment”. Risk Analysis, Volume 37, Issue 8, August 2017, Pages: 1495-1507.

Page 40: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

40

1) Video Image based Monitoring of Rail

2) Deep Neural Networks for Rail Monitoring

3) Making use of the data: Risk analysis

4) Conclusions

Outline

Page 41: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

41

Conclusions

• “Fancy” algorithms will not perform 100% if the knowledge of the railway system is not included explicitly.

• Purely data-based methods do not guarantee physical meaning. A combined approach, data-based with physical modelling would be preferred.

• There is a great potential for using Deep Learning to facilitate maintenance decisions on Dutch railways. Further research: head-checks, corrugation, wheel-burns, indentations.

Page 42: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

42

Conclusions

• Self-learning, transfer learning and new architectures could be tested.

• Higher resolutions cameras, including 3D measurements, can allow a complete digitalization of the railways assets.

• Many open challenges: Fusion of data, velocity, etc.

Page 43: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

43

A. Jamshidi, S. Hajizadeh, Z. Su, M. Naeimi, A. Núñez, R. Dollevoet, B. De Schutter and Z. Li, “A decisionsupport approach for condition-based maintenance of rails based on big data analysis”. Transportation

Research Part C: Emerging Technologies, Volume 95, October 2018, Pages: 185-206.

Video and ABA to detect squats.

Other signals for influential factors used for modelling.

How to keep improving?

Page 44: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

44

Deep learning: learned features

Just for fun:

https://affinelayer.com/pixsrv/index.html

https://playground.tensorflow.org/

Page 45: Automatic detection of rail surface defects using … faculteit...Microsoft PowerPoint - 2019 05 10 ANunez CiTG-ES Monitoring Community Author anunezvicencio Created Date 5/10/2019

45

Lunch lecture series of Monitoring Community10th May 2019

Automatic detection of rail surface defects using

video image: A case study in the Dutch railways

Dr. Alfredo Núñez VicencioAssistant Professor, TU Delft

Section of Railway Engineering