3. What is maintenance ?
Technical maintenance is intended to maintain or
improve the health of some asset. It forms an integral
part of any Asset health management strategy.
From Wikipedia, the free encyclopedia
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME3
4. Types of maintenance :
Corrective maintenance (CM)
Preventive maintenance (PM)
Predictive maintenance (PdM)
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME4
5. SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME5
Maintenance Strategies Methods Drawbacks
CM No planning is needed Failure is unpredictable
Increase down-time
High cost
Can affect quality of production
PM Time-based planning
Parts replaced before end of life
Increases cost
More investment
PdM Condition monitoring
Problem with components can be
identified prior to failure
Short-term investment is required
to purchase a predictive solution
Need machine learning
Need qualified team
Need a Data Scientist’s
6. What is Machine Learning (ML) ?
Machine learning is predicting future based on past
data
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME6
7. What is machine-to-machine (M2M) and the Internet of Things (IoT)?
Machine to machine (M2M) is a new concept that can be used to
describe any technology that enables networked devices to exchange
information and perform actions without the manual assistance of
humans.
The Internet of Things (IoT) is an other side of M2M
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME7
8. Key components of an M2M and IoT system include sensors, RFID, a
Wi-Fi or cellular communications link
M2M communication and IoT are used in telemetry, monitoring and
smart predictive maintenance solution
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME8
9. Predictive maintenance (also known as "forecast maintenance") is based on
the collection of field data to predict failure, it allows to anticipate
breakdowns and to schedule interventions that avoid costly downtime.
In order to refine the precision of predictions more and more, predictive
maintenance is no longer based on probability but on intelligent monitoring
by placing the devices under continuous surveillance in order to act just in
time before a malfunction.
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME9
10. Predictive maintenance is still poorly exploited, as maintenance managers
avoid machine breakdowns direct an exaggerated preventive maintenance
which consists in multiplying the interventions upstream in order to prevent
incidents that may never happen, knowing that this too cautious preventive
maintenance will also impose planned production shutdowns for preventive
maintenance actions with or without replacement parts.
For efficient maintenance, it is interesting to have a balanced ratio between
predictive and preventive and integrate predictive technologies into the
maintenance strategy to ensure the reliability of equipment and maintain a
good up-time.
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME10
11. Some methods such as vibration analysis, thermography and acoustic
measurement are very rich with the data collected, using the real-
time analysis algorithms, the data collected by these sensors can be
used to predict breakdowns, They do not happen and cause a
technician to act before being confronted with a problem.
Therefore, effective predictive maintenance will depend on the
reliability of the data collected and the analysis of this data and
powerful analytical tools such a machine learning.
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME11
12. Azure Machine Learning from Microsoft will make it easy to develop
and implement a predictive maintenance system based on artificial
intelligence and to revolutionize the industry.
Microsoft Azure allows you to quickly deploy your predictive
maintenance system with Microsoft Azure Machine Learning and
Microsoft Azure IoT Suite without any prior investment because it is a
cloud-based solution and you only pay for what you use.
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME12
13. Our Solution is a low-cost compact fanless DIN-rail embedded systems
are suitable for intelligent smart predictive maintenance solutions in
mission critical or harsh industrial environments and access multiple
data from sensors in real time to predict asset failure can avoid costly
downtime and reduce maintenance costs.
Driven by Azure Machine Learning (ML) from Microsoft, these
solutions detect even minor anomalies and failure patterns
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME13
14. SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME14
15. SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME15
16. SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME16
17. SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME17
18. We compare our solution with the standard and traditional solutions
of prediction in the market we find that:
Standard Supervisory Control and Data Acquisition (SCADA) solutions
are costly compared to the proposed solution with Microsoft
technology coupled with connected IoT objects and does not require
a prior investment for purchasing servers and building data-centers
for storage and analyze data.
Security : Need to improve SCADA security
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME18
19. Why we chose Azure Machine Learning (ML) to build our predictive maintenance solution ?
Predictive cloud-based solutions are available from various vendors, including Amazon, IBM,
and Google but Union Telecom thinks Microsoft Azure might be the best choice for many
raisons:
- it easy to configure Azure ML with minimum training without coding by using standard
predictive models or create your own models with R programming or using Phyton scripts.
- Scalability and flexibility : Ability to Scale on Demand
- Cost benefits and pricing model : Cost Competitive
- Support resources
- Security : Azure ML was built based on the Microsoft Security Development Lifecycle
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME19
20. Advantages of our predictive maintenance solution :
• Predict where, when, and why asset failures are likely to occur.
• Quickly identify primary variables as part of root-cause analysis
process.
• Optimize spare-parts inventory to reduce inventory costs
• Reduce maintenance operations costs
• Security of data is based on the Microsoft Security Development
Lifecycle.
• Scalability and flexibility
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME20
21. Our predictive maintenance solution can be used with minor
adaptations in :
- Remote Monitoring Systems for Solar PV Power Plant
- Remote Monitoring Systems for Connected car with OBD-2 adapters
- Remote Monitoring Systems for Wind Power Plant for Fault
Detection in Wind Turbine
- Remote Monitoring Systems for agricultural environment and
greenhouse
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME21
22. Thank you for your attention
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME22
23. Union Telecom helps customers to develop and integrate all the
software layers from the hardware to the cloud.
For further information :
Please contact
larbi.ouiyzme@uniontelecom.ma
www.uniontelecom.ma
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME23
24. Union Telecom, Embedded System Integrator
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME24
25. LinkedIn Bibliography : article published on November 22, 2016
La maintenance prédictive avec Microsoft Azure
https://goo.gl/2Xtylo
SEIT’17 – Internet of Things: Recent Innovations and challenges
Larbi OUIYZME25