[tutorial] building machine learning models for predictive maintenance applications - yan zhang

30

Upload: papisio

Post on 16-Apr-2017

8.887 views

Category:

Data & Analytics


2 download

TRANSCRIPT

Page 1: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang
Page 2: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

Sample ScenarioPredictive maintenance in IoT applications vs. traditional predictive maintenance concepts

Predictive problem: “When an in-service machine will fail?”

Machine learning approach

Problem formulation

Use caseInput data – publicly available aircraft engine run-to-failure data

Data labeling and feature engineering

Tools to build end-to-end solution from data to web serviceAzure ML

Predictive Maintenance Template in Azure ML

Demo: desktop app to predict machine’s remaining useful life

2

Page 3: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

3

Page 4: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

DATA

Business apps

Custom apps

Sensors and devices

ACTION

People

Automated Systems

Data Science Process

DATA

4

Page 5: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

DATADesktop app

5

Page 6: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

6

Predictive Maintenance in IoT Traditional Predicative Maintenance

GoalImprove production and/or maintenance

efficiency

Ensure the reliability of machine

operation

DataData stream (time varying features), Multiple

data sourcesVery limited time varying features

Scope Component level, System level Parts level

Approach Data driven Model driven

Tasks

Failure prediction, fault/failure detection &

diagnosis, maintenance actions

recommendation, etc. Essentially any task

that improves production/maintenance

efficiency

Failure prediction (prognosis),

fault/failure detection & diagnosis

(diagnosis)

Page 7: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

7

1 1 5 4 3

7 5 3 5 3

5 5 9 0 6

3 5 2 0 0

Page 8: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

8

Page 9: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

9

Page 10: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

http://ti.arc.nasa.gov/tech/dash/pcoe/prognostic-data-repository/

10

Page 11: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

Sample training data~20k rows,

100 unique engine id

Sample testing data~13k rows,

100 unique engine id

Sample ground truth data100 rows

11

Page 12: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

12

Page 13: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

Sample training data~20k rows,

100 unique engine id

Sample testing data~13k rows,

100 unique engine id

Sample ground truth data100 rows

13

RUL label1 label2

?

Page 14: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

14

id cycle … RUL label1 label2

1 1 191 0 0

1 2 190 0 0

1 3 189 0 0

1 4 188 0 0

… … … …

1 160 32 0 0

1 161 31 0 0

1 162 30 1 1

1 163 29 1 1

1 164 28 1 1

1 165 27 1 1

1 166 26 1 1

1 167 25 1 1

1 168 24 1 1

1 169 23 1 1

1 170 22 1 1

1 171 21 1 1

1 172 20 1 1

1 173 19 1 1

1 174 18 1 1

1 175 17 1 1

1 176 16 1 1

1 177 15 1 2

1 178 14 1 2

1 179 13 1 2

1 180 12 1 2

1 181 11 1 2

1 182 10 1 2

1 183 9 1 2

1 184 8 1 2

1 185 7 1 2

1 186 6 1 2

1 187 5 1 2

1 188 4 1 2

1 189 3 1 2

1 190 2 1 2

1 191 1 1 2

1 192 0 1 2

Predefined window size

for classification models

w1 = 30

w0 = 15

w1

w0

Page 15: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

15

Page 16: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

16

a1 a2 … a21 sd1 sd2 … sd21 RUL label1 label2

Other potential features: change from initial value, velocity of change, frequency count over a

predefined threshold

Page 17: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

http://gallery.azureml.net (search “predictive maintenance”)

Page 18: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

http://azure.com/mlfree tier & standard tier

18

Accessible through a web browser, no software to install

Best ML algorithms

Extensible, support for R & Python

Collaborative work with anyone, anywhere via Azure workspace

Visual composition with end2end support for data science workflow

Page 19: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

19

Step #2B

Train and evaluate binary

classification models

Step #1 Data preparation and

feature engineering

Step #2A

Train and evaluate regression

models

Step #3A

Deploy web service with a

regression model

Step #3B

Deploy web service with a

binary classification model

Step #3C

Deploy web service with a

multi-class classification

model

Step #2C

Train and evaluate multi-class

classification models

Step 1 Step 2 Step 3

Page 20: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

20

Page 21: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

21

Page 22: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

22

Step #2B

Train and evaluate

binary classification

models

Step #1 Data

preparation and

feature engineering

Step #2A

Train and evaluate

regression models

Step #3A

Deploy web service

with a regression

model

Step #3B

Deploy web service

with a binary

classification model

Step #3C

Deploy web service

with a multi-class

classification model

Step #2C

Train and evaluate

multi-class

classification

models

Page 23: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

23

Decision Forest Regression

Boosted Decision Tree Regression

Poisson Regression

Neural Network Regression

Page 24: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

24

Saved Transform

Web service input/output

Page 25: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

25

Page 26: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang
Page 27: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

Desktop app

Azure ML Model (Deployed Web

Service)

ML predictions consumed through the RRS web service interfaceData input

27

Page 28: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

DATA INTELLIGENCE ACTION

28

Page 29: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

using three machine learning models: regression, binary classification, multi-class classification

Introduced how to build end-to-end data pipeline with Azure ML

29

Page 30: [Tutorial] building machine learning models for predictive maintenance applications - Yan Zhang

Microsoft Azure Machine Learninghttp://azure.com/ml

http://gallery.azureml.net (search “predictive maintenance”)

Register for the Cortana Analytics Workshop hosted in Redmond on September 10-11, 2015. https://analyticsworkshop.azurewebsites.net