Predictive Maintenance using MATLAB: Pattern Matching for ... · Predictive Maintenance using...

Dr. Türck Ingenieurbüro

Data Science

Predictive Maintenance using MATLAB:

Pattern Matching for Time Series DataDr. Irina Ostapenko (Dr. Türck Ingenieurbüro GmbH),

Jessica Fisch (Daimler AG)

26.06.2018

Dr. Türck IngenieurbüroZwischen den Zahlen lesen

www.mercedes-benz.com

www.tuerck-optik.de

Data Science Predictive Maintenance using MATLAB: Pattern Matching for Time Series Data 2

Irina Ostapenko

Senior Data Scientist

Solutions and Algorithms

Kreuzbergstr. 37 Berlin 10965

io@tuerck-optik.de

Jessica Fisch

Mercedes-Benz Werk Mettingen

Digitale Fabrik Powertrain und Projekt Industrie 4.0

PT/TSD

010 – HPC M427

70546 Stuttgart

jessica.fisch@daimler.com

Mercedes-Benz

Collaboration partners

Data Science Predictive Maintenance using MATLAB: Pattern Matching for Time Series Data 3

Irina Ostapenko

io@tuerck-optik.de

Jessica Fisch

Collaboration partners

We provide

• Algorithms

• Signal Processing

• Measurement Systems Developing

Optical System Design

Mercedes-Benz

Focus:

• Digital Transformation

• Big Data

• IIoT

Data Science

Outline

1. Project introduction

2. Task description

3. Solution/Algorithm

4. Summary

Predictive Maintenance using MATLAB: Pattern Matching for Time Series Data 4

Data Science

POWERTRAINFive modules form

the core of our cars

Data Science

The Powertrain production network is set up globally

with lead plant in Germany

Locations at a glance

Europe USA

Beijing

(Joint Venture with

BAIC Motor)

Decherd

(Cooperation with

Renault/Nissan)

Legend

100% Daimler Joint Venture

Majority Holding Cooperation

Berlin

Hamburg

Koelleda (MDC Power)

Arnstadt (MDC Technology)

Stuttgart

Rastatt

Jawor/Poland

Most/ Czech Republic

Maribor/Slovenia

Sebes and Cugir/Romania

(Star Transmission)

Kamenz (ACCUmotive)

Data Science

Motivation for Anomaly Detection in the Projekt „iLL“

Source: www.autem.de

Automatic Notification of the Deviation:PLC-Data today:

Data properties in the context of Big Data

The 3 basic V's of Big Data:

• Velocity: Speed with which data is

generated and analyzed

• Volume: Amount of data that traditionally

can not be analyzed

• Variety: Data diversity refers to

unstructured data without a recognizable

context

The 2 additional V‘s:

• Validity: Ensuring data quality

• Value: measurable benefits from the data

Workshop zur Fabrik des Jahres | MB Werk Berlin | 05.03.18 8

Volume VarietyVelocity

sampling rate ≈100 ms

700 Variables6 TB/p.M.

Maschine

Machine State

Numerical Control

Big Data

2 GB/day

100 Machine =

Data Science

Machine

Benefits of „Intelligent Level-Learning“

Bearing

damaged !

Different cycle× Cycletime

× power

✓ Position

… …

--- Active power engine axis 1

--- Position axis 1

Normal cycle✓ Cycletime

✓ power

✓ Position

Error cycle× Cycletime

× power

× Position

Repair

Procurement

of spare

parts;

Downtime in

production

Create

maintenance

order for

unplanned

breakdown

Ensuring lean production by

controlling quality, cost and time

Data Science

Challenges

• About 700 parameters are continuously monitored in every production cycle

yielding 700 individual time-series of about 2500 samples each

Time, s

Data Science

Challenges

• Different parameters show very different and elaborate features

Time, s

Data Science

Challenges

• Different parameters show very different and elaborate features

Time, s

Task: Analyse these 700 time-series and find specific kinds of deviations

Data Science

Requirements for algorithm

Time, s

Time deviation

Data Science

Time, s

Time deviation

Time, s

Pattern deviation

Data Science

Time, s

Time deviation

Time, s

Pattern deviationmight not be critical is critical

Data Science

What the algorithm should do

• Time series analysis

• Find deviations from normal cycle and

• Distinguishing between time and pattern deviation What is normal?

Time, s

Time deviation

Time, s

Pattern deviationmight not be critical is critical

Data Science

Delays in production cycles

• Length of time-series varies from cycle to cycle

even for normal production

Cycle lengths over one day, s

lative

Data Science

Time, s

lative

1 cycle

Data Science

Time, s

lative

2 cycle

Data Science

Time, s

lative

4 cycle

Data Science

Time, s

different delays

lative

Cycles from one day

Data Science

Time, s

perfect match

after shift in time axis

Time, s

different delays

lative

Normal cycles can be matched to one another

through shifting in time axis!

Cycles from one day

Data Science

Algorithm principle

rmal c

Data Science

Algorithm principle

rmal c

• Cycle can be described as sequence of

features f1, f2, f3

• Each cycle can show some delays in

time t1, t2

Data Science

Algorithm principle

rmal c

features f1, f2, f3

time t1, t2

• Automatic feature detection f1, f2, f3

Data Science

Algorithm principle

rmal c

Δt3Δ t1 Δ t2

features f1, f2, f3

time t1, t2

• Pattern matching through shift of feature

along time axis (Δt2, Δt2, Δt3):

least square fit (tshift to minimize the Sum

of Residual Squares of two signals)

• Automatic feature detection f1, f2, f3

Data Science

Algorithm principle

Δt3Δ t1 Δ t2pattern deviation

time deviation

• Pattern matching through shift of feature

along time axis (Δt2, Δt2, Δt3):

minimization of SRS

Data Science

Algorithm principle

Δt3Δ t1 Δ t2pattern deviation

time deviation

• Pattern matching through shift of

feature along time axis (Δt1,

Δt2, Δt3): minimization of SRS

• Time and pattern deviation

for each feature are used

as characteristic numbers for test cycle

f1 f2 f3

Δt1 Δt2 Δt3 Time deviation

No No Yes Pattern deviation

Data reduction!

• Decription of a cycle as feature

sequence

• For each feature time and pattern

deviation can be calculated

Data Science

Automatic feature detection

Time series is split

• After a local extremum (maximum or minimum) or on a plateau

• After a given relative change

Time, s

4 features 7 features 14 features

Data reduction of time series from 2500 datapoints to sequence of max. 60 features (typically 10)!

Data Science

Algorithm implementation: machine learning approach in MATLAB

Training

Reference cycles ->

• Build „reference signal“ for each feature

• Limits for time and pattern deviation

Testing

Test cycles ->

• Comparison of each feature in reference signal

• Is time and pattern deviation within the limits?

Data Science

Create „reference signal“ for each produciton parameter

1. For all training cycles - matching to shortest cycle

2. Create „reference signal“ – mean over all matched reference cycles

Time, s

Almost perfect match after shifting

Cycles from one day

Time, s

Reference signal

Training

Data Science

1. For all training cycles - matching to shortest cycle

3. Possible pattern deviation - standard deviation over all matched reference cycles

Time, s

Almost perfect match after shifting

Cycles from one day

Time, s

Tolerance for pattern deviation

Time, s

Reference signal

Training

Data Science

2. Possible pattern deviation - standard deviation over all matched reference cycles, limits for SRS

3. Possible time deviation – maximal absolute shift from matched reference cycles

Nr. of feature

Time, s

Tolerance for pattern deviation

Nr. of feature

Possible time deviation

Possible pattern deviation

Training

Data Science

Testing: time and pattern deviation evaluation

error cycle

normal cycle

reference signal

Time, s

Testing

Data Science

error cycle

normal cycle

reference signal

Time, s

Is time and pattern deviation

for this feature within the limits?

• Tolerance window (Δ t, SRS)

• Easy to spot a critical deviation

Testing

Time: max Shift (norm.)F

feature a

Data Science

error cycle

normal cycle

reference signal

Time, s

dTesting

max Shift (norm.)

feature d

max Shift (norm.)m

feature c

max Shift (norm.)

feature b

max Shift (norm.)

feature a

Data Science

Algorithm: Summary

1. Quantitative and qualitative description of production failure

2. Independent of signal form -> universally applicable to other applications or machines

3. Signal description with characteristic numbers, which are easy to interpret

4. Data reduction with a factor 250 without significant loss of information!

5. Easy control of production: recognition of critical errors and non-critical delays online

io@tuerck-optik.de

Data Science

Example of the added value

Results:

• Transparency of the process

• Deviation for each Signal

• Reason of Cycletime increase found

Time and Pattern deviation are recognized

Example: Change in code

normal

actual

deviation

0 50 100 150 200 250

Position Setpoint Spindle Axis W1

Time [s]

Data Science

Summary

Algorithm using pattern matching for time series developed and implemented for production data

Why MATLAB?

• easy algorithm implementation

• existing solution for data import

• very good support and broad use in universities

MathWorks products used:

• Signal Processing Toolbox

• Statistics and Machine Learning Toolbox

Outlook:• Parallel Computing Toolbox for performance improvement

Prototyp intelligent Level-Learning (iLL) has a new function for anomaly detection• Troubleshooting in case of failure (maintenance), Parts Planning, Influences on the quality

Optimization of repair time, spreaders amount, …

Data Science

Thank you for your attention!

Irina Ostapenko

io@tuerck-optik.de

Jessica Fisch

jessica.fisch@daimler.com

Mercedes-Benz

Predictive Maintenance using MATLAB: Pattern Matching for ... · Predictive Maintenance using...

Documents

Java Design Pattern

A MATCHING STORY - WordPress.com€¦ · das Medizin-StartUp: Eine unerwartete Matching-Story „Es wird sich schon etwas auftun“ – „Mit Geisteswissenschaften kannst du alles

Datengetriebene Module für Predictive Maintenancedbis.eprints.uni-ulm.de/1491/1/...Predictive-Maintenance_productivity17... · Prädiktive Instandhaltung 21 Das Themengebiet Predictive

String Matching - Komplexe Informationssysteme · String Matching Komplexe Informationssysteme Fabian Panse panse@informatik.uni-hamburg.de Universit at Hamburg

Entwurfsmuster (Patterns) Prof. Dr. Matthias H olzl Joschka Rinke · Beispiel: Design-Pattern im GoF-Katalog Beschreibungsform f ur Design Pattern I Pattern Name and Classi cation

«Predictive Leadership – future now»

Predictive Linkbuilding

Matching mit den vorhandenen Anwendungsroutinen - SAS-Wiki

StickyPillars: Robust and Efficient Feature Matching on

Hardware Pattern Matching for Network Traffic … · Technische Universität M unchen¨ Fakultät f ur Informatik¨ Diplomarbeit in Informatik Hardware Pattern Matching for Network

Graph Pattern Semantik

Tattered Lace Pattern Collection 1 Patronencollectie Tattered Lace … · 2019-04-19 · E N Tattered Lace Pattern Collection 1 Pattern List N L Patronencollectie Tattered Lace 1

12c SQL Pattern Matching wann werde ich das benutzen Andrej Pashchenko

Graph Matching Torsten Gründel 03.11.2006. Überblick 1. Was ist Graph Matching 2. Morphismen 1. Allgeimeines 2. Graphisomorphismus 3. Subgraphisomorphismus

SAP Predictive Analysis 1.18

Alphabet Letter Matching, 000 oo 123kidsfun123kidsfun.com/images/pdf/easter/123kidsfuncom_easter... · 2018-03-09 · Alphabet Letter Matching, 000 oo 123kidsfun.com . Created Date:

Big Data und Predictive Ana ly tics & Digitale ... · Big Data und Predictive Ana ly ... zur Realisierung von Predictive Analytics und Pre-dictive Maintenance vorgestellt. So können

Profiling – Matching - Vermittlung. Inhalt Übersicht Bewerber und Stellen Matching Vermittlung Web-Grabber Kontaktmanagement Online-Börse Infosystem

Statistisches Matching – Anwendungsmöglichkeiten ... · PDF fileZA-Information 51 1 entnommen aus: Bacher, J., 2002: Statistisches Matching: Anwendungsmöglichkeiten, Ver-fahren

Digital Drum Bedienungsanleitung - musicstore.de€¦ · Mit Pattern üben Ein Pattern ... 10. DRUM OFF Taste Schalten Sie das Schlagzeug im Pattern stumm. 11. TEMPO Taste Stellen