Silent intelligence in manufacturing industry · DATA LAKE (AGGREGATED RAW DATA IN MULTIPLE FORMATS) EXTERNAL ... CONTROL AND OPTIMIZATION THROUGH, FOR EXAMPLE, SOFTWARE UPGRADES

Silent intelligencein manufacturing industry

Bhoopathi RapoluHead of Analytics – EMEA | CYIENT Europe Ltd.

CY

IEN

T©

20

15

3/1

2/2

01

5

How Smart Connected Products are TransformingManufacturing Industry

Digital Technologies for Manufacturing Innovation: Embracing Industry 4.0The University of Nottingham | 30 November 2015

2

CYIENT – Who we are

Aerospace Heavy Equipment Oil & Gas

Medical & ElectronicsRail Transportation Utilities & Energy

CY

IEN

T©

20

15

3/1

2/2

01

5

3

CYIENT – What we do

Advanced AnalyticsIot And M2mBig Data

Product realizationEngineering design Network & Operations

CY

IEN

T©

20

15

3/1

2/2

01

5

Silent intelligence

in manufacturing industry

How Smart Connected Products are TransformingManufacturing Industry

5

The New Product Anatomy

• Sensors• Microprocessors• Data storage• Controls• Software• Embedded OS• Digital UI etc.

• Ports• Antennae• Protocols• Networks

• Mechanical parts• Electrical and Electronic

parts

ConnectivityComponents

6

Reinventing the Product Design

MechanicalEngineeringMechanicalEngineering

Interdisciplinary SystemsEngineering

Interdisciplinary SystemsEngineering

Monitor&

Report

Monitor&

Report

Enableremotecontrol

Enableremotecontrol OptimiseOptimise

SoftwareDefinedSoftwareDefined Auto-

nomousAuto-

nomous

EfficientEfficient

NewBusinessModels

Interoperability

SimplifiedAssemblyprocess

Insights

Optimiseproduction

7

BUSINESS CUSTOMER PARTNER

DIAGNOSTIC DESCRIPTIVE PREDICTIVEPRESCRIPTIV

ECapture products’condition,environment, andoperation

Examine the causesof reduced productPerformance or failure

detect patterns thatsignal impendingevents

Identify measuresto improve outcomesor correct problems

DATA LAKE(AGGREGATED RAW DATA

IN MULTIPLE FORMATS)

DATA LAKE(AGGREGATED RAW DATA

IN MULTIPLE FORMATS)

EXTERNALdata on prices, weather,supplier inventory, etc.

SMART, CONNECTEDPRODUCTS

data on location,condition, use etc.

ENTERPRISEdata on service histories,

warranty status, etc.

CONTROL AND OPTIMIZATIONTHROUGH, FOR EXAMPLE,

SOFTWAREUPGRADES THAT IMPROVE

PERFORMANCE

BASIC INSIGHTSSUCH AS USE

PATTERNS

Unlocking the full value of data is a new business function

Source: HBR article on smart connected products, Oct 2015

Cyient case studies

Observations:1) Are outliers present in data?2) What are the common

characteristics of these outliers?3) What is their impact on overall

performance objectives?

Method:

Hierarchical clustering and predictivemodelling

Output:1) TRANSIENT behavior –

Are componentsdynamically changing clusters?

2) Contributing REASONS - Why, ifso?3) FEATURE extraction - What arethe features that are contributing tosuch migration?

Airframe OEM: Prescriptive Analytics

10

Condition BasedMonitoring Module

Situational

Intelligence Engine

Deduction ofRisk categories Recommendations

SituationalIntelligence

OperationalIntelligence

NewInsights

• Anomaly detection• Similar Asset based analysis• Fleet based analysis• Site based analysis

• Asset Health Score• In-between repair time estimation• Important component failure prediction

• Repair Recurrence models• Event – Repair Association• Risk category to health quantification

Customer Benefits

Sensor info, abuse events,exceptions, fluid samples,repair history, site conditions.– Ove 170 different data sets

1000

D

D

F

F

C

C

B

B

A

AA

A A AB

B

B

C

D

F

F

A

A

B

B

C

D

F

Time

He

alth

Sco

re

0

Estimated Life

Heavy Equipment Fleet – Predictive Maintenance

CY

IEN

T©

20

15

3/1

2/2

01

5

11

Imaging system MR imaging controls

Onboard sensors Captures temperature at various

positions on MR Machine

Other data sourcesEquipment HistoryMaintenance RecordsEnvironmental data

FailureEvents

PatternMatchingAlgorithm

Value to Client

− More accurate prediction of failures in MR machines− Temporal pattern recognition across failures − Efficient scheduling of maintenance visits to client installations− Better predictive maintenance of MRI machines− Optimized inventory holdings at Field Service Locations

Magnetic Field Control Manual setting of magnetic

field required for scan asprescribed by the doctor

(DAS) DataAcquisition System

SystemFailureLogs

Operations(Scan) Data

Patient Data Body part for scan Weight of patient

System HealthDashboard

Predictive AssetMaintenance

Predictive Analytics Platform

Predictive Maintenance for MRI MachinesC

YIE

NT

©2

01

53

/12

/20

15

Thank You

Bhoopathi RapoluHead of Analytics – EMEA | CYIENT Europe Ltd.

www.Bhoopathi.comTwitter: @bhoopathi

CY

IEN

T©

20

15

3/1

2/2

01

5