You are on page 1of 15

Digital Factory

Robot Predictive
maintenance
Why use predictive maintenance?
Predictive maintenance employs
Spot degradation early with
machine learning to automatically predictive maintenance
spot anomalies in asset health in real-
time
Early identification of degradation
prevents serious failure and downtime
Learning from each equipment can
be used to diagnose health of all
similar equipment
Spares inventory, service schedules
and asset replacement can be Asset degradation stages
optimized and planned in advance
Machine learning steps
Build and train
Problem ML model
definition

Choose ML Evaluate ML
Data sourcing Deploy model
algorithm model
into production

Data
preparation Visualize and
refine

Machine Learning model development


Determine optimum settings for desired
output
Predicted
Input variables Output variables
• Temperature,
Humidity,
• Vibration, Current
• Line Pressure • Brake Wear
• Speed / Torque / Time • Lifetime
• Dynamic behavior Trained • Energy consumption
• Material Consumption model • Efficiency
• Waste Material / • …
Metals
• Relational
Parameters eg: load,
gear etc
• …

Apply various “What If” scenarios with different input variable combinations and examine likely outcomes
System architecture
Dashboards, reports and Export to
Apps
alerts Maximo, SAP,…

Trained Prediction
PM Application
model engine

Data Archive Time-series Unstructured Relational


data data data

PLC Gateway

Context Failure,
Robots Sensors
data downtime data
Servo Amplifier temperature,
Servomotor temperature /
Current, Brake Wear, Gear
Reduction boxes temperature,

Output
Asset health score
Maintenance schedule
Top failure reasons
Failure probability
Likely failure parts

Predictive Analytics benefits

manufacturing

quality control

health, safety,
procurement

maintenance

analyst/ data
engineering/

engineering/

environment
production/

statistician/
operations

finance &
reliability

warranty
planning

scientist
service/
product
process

design
reduce unplanned downtime, improve asset reliability, availability
& utilization
     

extend asset life; avoid new asset costs    


maintain or improve product quality     
reduce scrap & process variability; improve yield      
improve manufacturing/production process       
reduce maintenance costs    
optimize maintenance strategies & practices   
optimize spare/critical parts inventory     
monitor asset health; remote diagnosis; root cause analysis; predict
pending failures
       

optimize supplies, materials, components inventory       


reduce warranty & service costs      

reduce risk & compliance exposure           

8
IMS - Integrated
Manufacturing System
System architecture
Mobile / tablet Web app – dashboards, Reports, Business systems
apps forms, … alerts integration

Automatic scheduler
ML models Common visualization platform
(Siemens)

OPC Real-time
FTP Historical
SNMP Enterprise gateway Big Data Aggregated
ODBC Relational
REST …
Flat file

FIS NGAVS MFM QLS SMART

Plant equipment – PLC, SCADA, Test station, meters, field gateway, …


Automated, closed-loop scheduling
Demand
ERP
Order
Amendments New
Order Update
Material
Availability Schedule
Analyse
SIMATIC IT Constraints Sequence Schedule
Preactor AS
Generate
Order Progress
Schedule
Breakdowns
Update Release

Execute
MES/SCADA
Addon - Virtual Commissioning
Automation
Validation
 Easily create a realistic shadow of
a production environment

Simulation
 Test & Debug PLC code
 Long before the real system is build
 Minimize impact on existing Real Time
environment Signal
Exchange
 Use real PLC H/W or emulated OPC (Client & Server)
S/W

Controls
PLC Code HMI

PLC
Addon - Plant simulation
Create a dynamic computer model
of a complex system to explore its
characteristics and optimize
performance
Run experiments and what-if
scenarios without disturbing
production
Minimize cost of new investment
while maintaining same output
IMS Features and Benefits
Central, consolidated data store for all plant data -> long-term experience gathering
Cross-discipline data availability for easy analysis, machine learning -> many
opportunities for continuous improvement
Integration of all production resources, constraints and processes for automated
scheduling -> higher output
Single-window analysis of throughput, efficiency, issues, root causes -> better
quality, reduced costs
Build intelligence with addon systems leveraging the common data and knowledge
base -> significant competitive edge
Thank you

You might also like