SlideShare a Scribd company logo
Speaker: Mauricio Fadel Argerich
NEC Laboratories Europe GmbH
Model Learning for Cloud-Edge
Digital Twin
MoLe: Model Learning for Cloud-Edge Digital Twin
• With MoLe we aim to simplify the
implementation and execution of Digital Twins
in digital factories by:
• Utilizing FIWARE technologies (NGSI-LD, Scorpio,
FogFlow) to dynamically orchestrate a setup to
generate a Digital Twin from data and execute it
on Edge and Cloud
• Reducing effort needed to develop prediction and
simulation models by using Knowledge Infusion
I4.0Lab – Manufacturing & Assembly process
• We have developed our solution for the MIDIH
Didactic Factory in Milan
• The factory implements the Manufacturing and
Assembly process of PCBs
• 7 steps, carried out by different stations
Translator and Scorpio
• Stations in didactic factory use NGSIv2, Scorpio
uses NGSI-LD
• Translator reads sensor data from the Factory
Information Bus and uses Kafka information (topic,
key, message schema) to transform it into NGSI-LD
entities
• Data in NGSI-LD is received by the Scorpio Broker,
where it is stored and sent to any subscribers
• Scorpio optimizations:
• We optimized data handling and serialization, as well as
error handling
• We optimized Scorpio’s vertical scaling and internal data
handling
Translator
DB
Subscriber
Subscriber
Subscriber
Subscriber
Scorpio
Results: Translator and Scorpio
Translator
• Worst case delay of 1ms between a message arrival
in the Kafka bus and the retrieval in the Python
library
• Below KPI of 2Hz (sensors sampling frequency)
Scorpio
• We made a comparison with the updated version of
the FIWARE GE Orion called Orion-LD
• Scorpio achieved in average half the latency of Orion
Scorpio
Orion
Digital Twin Models
• We implemented Digital Twins (DTs) for the
different stations in the M&A process
• DTs are programmable objects that can be
instantiated for real-time monitoring and
simulations
• DTs of Front Cover Magazine and the Press Station
• These DTs implement specific models to:
• detect the current status of the station based on its
current sensor and actuators data
• predict energy usage based on the same data
Knowledge Infusion
• Energy usage prediction à pure ML
• State inference model à Knowledge Infusion (KI) = ML + domain knowledge
• Domain knowledge is infused through Knowledge Functions (KFs)
• KFs output a single value: a label.
• Functions implement human provided logic and utilize facts derived from internal and
external knowledge bases
• Types of knowledge functions: Weak and Strong
• KI creates a Knowledge Model that serves two purposes:
• Data augmentation: improves data quality by creating new features or labelling data
• Robustness: allows us to correct some obviously wrong outputs of the ML model during runtime
ML
Model
Knowledge
Model
KI Model
FogFlow
• FogFlow was extended to serve ML models as serverless fog functions
• ML models are implemented based on FogFlow ML operators
• To implement a FogFlow ML model, we follow 3 steps:
1. Model registration: to register a ML through a web-based GUI
2. Model deployment: to create and deploy a serverless fog function in FogFlow to run the ML model
3. Model serving: to apply the ML model inside the deployed function instances to produce the
detection/prediction result of the input data and then update the state of the corresponding DT
Results: Digital Twin Models
• Model for the station’s state
• Small high quality dataset: manually labeled 10m of
data (2Hz, 1200 data points total) as “Idle” or
“Working”
• Larger noisy dataset: around 50m or 6000 data points,
labelled with 2 simple functions
• We used the small dataset to train a Random Forest
Classifier (RFC) using all the features (45)
• First 600 samples to train RFC and last 600 to evaluate it
• The RFC achieved an accuracy of 82.11% on its test set
• We also implemented a Knowledge Model (KM),
based on two programmable functions
• Each function took no more than 5 minutes, they check
variable values and return state of the machine
• Test accuracy is 70%
Performance of RFC on test set when trained
with 50% of manually labeled data
Performance of KM on test set with 2 simple
labelling functions
S
L
train
test
à
Results: KI for Digital Twin Models
• We can utilize KM to label larger unlabeled dataset and
train a ML model with it
• Hopefully, the ML will learn to filter out the noise
• The ML model is trained with 10x more data as
before
• We re-trained the RFC with these larger noisy dataset
• Slight improvement in its performance: 82.33%
• Without any manually and costly labeled data!
• KM took us about 10 minutes to implement
• We have also implemented a KI model that utilizes a
supervision function
• This function verifies the value of certain variables
and forces the output of the ML model
• This function represents a layer of safety
• Accuracy in our tests remained the same
Results: Energy Consumption Prediction
• We implemented a Random Forest
Regressor to deal with correlated and non-
informative variables
• Energy consumption is influenced by activity of
the station, logged by its sensors
• There are variables which are correlated and
others are nearly constant throughout the
activity, it seems we might have partial visibility
• We trained it using 80% of the full time
series data for Press Station and kept the
20% remaining as test set
• The Random Forest Regressor obtained a Mean
Absolute Error of 2.75. train
test
à
Results on test set
KPIs
KPI 1: Data velocity
1000 msgs per second >> 2 msgs per second (sampling frequency of
sensors in digital factory)
KPI 2: Generation of knowledge graph for the DT
The Translator is capable of generating a graph structure in the form
of NGSI-LD Entities. This is done automatically using provided meta
data from the factory message bus.
KPI 3: Accuracy of KI for DT
Accuracy of ML model trained with hand labeled data: 82.11%
Accuracy of Knowledge Model: 70.50%
Accuracy of ML model trained with data labeled by Knowledge
Model: 82.33%
KPI 4: Accuracy of DT Refinement (Strong KF)
Accuracy of KI model with strong KF: 82.33%
Note: Same accuracy as without refinement because Strong KF did
not find any necessary correction
Lessons learnt
NGSI Translator: Valuable tool for extracting knowledge from raw data, allows for more flexibility
(NGSIv2, NGSI-LD). You can find the translator at
https://github.com/ScorpioBroker/ScorpioBroker/tree/feature-82/NGSILDTools/NGSILDTranslator
KAFKA NGSI-LD Integration: Kafka is a good choice as it is wide spread and has excellent performance.
Recommendation: use key on Kafka messages so an identifier is attached to the data.
Digital Factories Data: Data heterogeneity between factories is still very high.
Opportunity: tools/techniques to join data from different factories are valuable and needed!
Recommendation: it’s beneficial to publish example data from MIDIH factories.
Knowledge Infusion: it enabled us to train a classifier with no manually labeled data, achieving high
accuracy. KI shows great potential to reduce effort of creating ML models.
KI, ML models’ performance and execution in FogFlow: Models achieved good accuracy but we
believe this can be further improved.
THANK
YOU!
NEC Laboratories Europe

More Related Content

What's hot

ME490B - SMART BOARD FINAL REPORT
ME490B - SMART BOARD FINAL REPORTME490B - SMART BOARD FINAL REPORT
ME490B - SMART BOARD FINAL REPORT
Gabriel Dariano
 
REMOTE TRIGGERED SOFTWARE DEFINED RADIO
REMOTE TRIGGERED SOFTWARE DEFINED RADIOREMOTE TRIGGERED SOFTWARE DEFINED RADIO
REMOTE TRIGGERED SOFTWARE DEFINED RADIO
Kunal Bidkar
 

What's hot (20)

ME490B - SMART BOARD FINAL REPORT
ME490B - SMART BOARD FINAL REPORTME490B - SMART BOARD FINAL REPORT
ME490B - SMART BOARD FINAL REPORT
 
REMOTE TRIGGERED SOFTWARE DEFINED RADIO
REMOTE TRIGGERED SOFTWARE DEFINED RADIOREMOTE TRIGGERED SOFTWARE DEFINED RADIO
REMOTE TRIGGERED SOFTWARE DEFINED RADIO
 
About Sigma ML | MCPS Inc
About Sigma ML | MCPS IncAbout Sigma ML | MCPS Inc
About Sigma ML | MCPS Inc
 
Historic Opportunities: Discover the Power of Ignition's Historian
Historic Opportunities: Discover the Power of Ignition's HistorianHistoric Opportunities: Discover the Power of Ignition's Historian
Historic Opportunities: Discover the Power of Ignition's Historian
 
Manage Infrastructure at Scale with Automation - June 2020
Manage Infrastructure at Scale with Automation - June 2020Manage Infrastructure at Scale with Automation - June 2020
Manage Infrastructure at Scale with Automation - June 2020
 
Performance Testing Internet of Things
Performance Testing Internet of ThingsPerformance Testing Internet of Things
Performance Testing Internet of Things
 
D02: Performance Engineering and Testing of Predix Apps (Predix Transform 2016)
D02: Performance Engineering and Testing of Predix Apps (Predix Transform 2016)D02: Performance Engineering and Testing of Predix Apps (Predix Transform 2016)
D02: Performance Engineering and Testing of Predix Apps (Predix Transform 2016)
 
PEM2: Control Applications Portfolio from GE Power
PEM2: Control Applications Portfolio from GE PowerPEM2: Control Applications Portfolio from GE Power
PEM2: Control Applications Portfolio from GE Power
 
Roofline Model for FPGA: A tool for Performance Analysis and Application Opti...
Roofline Model for FPGA: A tool for Performance Analysis and Application Opti...Roofline Model for FPGA: A tool for Performance Analysis and Application Opti...
Roofline Model for FPGA: A tool for Performance Analysis and Application Opti...
 
Affordably Refreshing Your Water District’s Process Control
Affordably Refreshing Your Water District’s Process ControlAffordably Refreshing Your Water District’s Process Control
Affordably Refreshing Your Water District’s Process Control
 
Learn how Intuit created an application-aware network performance platform
Learn how Intuit created an application-aware network performance platformLearn how Intuit created an application-aware network performance platform
Learn how Intuit created an application-aware network performance platform
 
Intelligent Digital Mesh Testing
Intelligent Digital Mesh TestingIntelligent Digital Mesh Testing
Intelligent Digital Mesh Testing
 
Automated testing in migration to 5G network and the cloud
Automated testing in migration to 5G network and the cloudAutomated testing in migration to 5G network and the cloud
Automated testing in migration to 5G network and the cloud
 
Siddharth more resume_obj_c
Siddharth more resume_obj_cSiddharth more resume_obj_c
Siddharth more resume_obj_c
 
SiPCPE-108-1
SiPCPE-108-1SiPCPE-108-1
SiPCPE-108-1
 
Complete low cost scada system of the intelligent house
Complete low cost scada system of the intelligent houseComplete low cost scada system of the intelligent house
Complete low cost scada system of the intelligent house
 
Open Daylight Forum India 2015
Open Daylight Forum India 2015Open Daylight Forum India 2015
Open Daylight Forum India 2015
 
Top 10 Design & Security Tips to Elevate Your SCADA System
Top 10 Design & Security Tips to Elevate Your SCADA SystemTop 10 Design & Security Tips to Elevate Your SCADA System
Top 10 Design & Security Tips to Elevate Your SCADA System
 
MIDIH D-Cube-INART presentation
MIDIH D-Cube-INART presentationMIDIH D-Cube-INART presentation
MIDIH D-Cube-INART presentation
 
rpastage
rpastagerpastage
rpastage
 

Similar to Mo.Le nec-midih_presentation_oc2

“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...
“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...
“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...
Edge AI and Vision Alliance
 
Slides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdf
Slides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdfSlides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdf
Slides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdf
vitm11
 
Real-Time Voice Actuation
Real-Time Voice ActuationReal-Time Voice Actuation
Real-Time Voice Actuation
Pragya Agrawal
 

Similar to Mo.Le nec-midih_presentation_oc2 (20)

IncQuery-D: Incremental Queries in the Cloud
IncQuery-D: Incremental Queries in the CloudIncQuery-D: Incremental Queries in the Cloud
IncQuery-D: Incremental Queries in the Cloud
 
Webinar: Começando seus trabalhos com Machine Learning utilizando ferramentas...
Webinar: Começando seus trabalhos com Machine Learning utilizando ferramentas...Webinar: Começando seus trabalhos com Machine Learning utilizando ferramentas...
Webinar: Começando seus trabalhos com Machine Learning utilizando ferramentas...
 
Accelerating Machine Learning on Databricks Runtime
Accelerating Machine Learning on Databricks RuntimeAccelerating Machine Learning on Databricks Runtime
Accelerating Machine Learning on Databricks Runtime
 
Legion - AI Runtime Platform
Legion -  AI Runtime PlatformLegion -  AI Runtime Platform
Legion - AI Runtime Platform
 
Strata parallel m-ml-ops_sept_2017
Strata parallel m-ml-ops_sept_2017Strata parallel m-ml-ops_sept_2017
Strata parallel m-ml-ops_sept_2017
 
World Artificial Intelligence Conference Shanghai 2018
World Artificial Intelligence Conference Shanghai 2018World Artificial Intelligence Conference Shanghai 2018
World Artificial Intelligence Conference Shanghai 2018
 
Lessons Learned Replatforming A Large Machine Learning Application To Apache ...
Lessons Learned Replatforming A Large Machine Learning Application To Apache ...Lessons Learned Replatforming A Large Machine Learning Application To Apache ...
Lessons Learned Replatforming A Large Machine Learning Application To Apache ...
 
Machine Learning for .NET Developers - ADC21
Machine Learning for .NET Developers - ADC21Machine Learning for .NET Developers - ADC21
Machine Learning for .NET Developers - ADC21
 
MLOps.pptx
MLOps.pptxMLOps.pptx
MLOps.pptx
 
Operationalizing Machine Learning: Serving ML Models
Operationalizing Machine Learning: Serving ML ModelsOperationalizing Machine Learning: Serving ML Models
Operationalizing Machine Learning: Serving ML Models
 
PyCon Sweden 2022 - Dowling - Serverless ML with Hopsworks.pdf
PyCon Sweden 2022 - Dowling - Serverless ML with Hopsworks.pdfPyCon Sweden 2022 - Dowling - Serverless ML with Hopsworks.pdf
PyCon Sweden 2022 - Dowling - Serverless ML with Hopsworks.pdf
 
Key projects in AI, ML and Generative AI
Key projects in AI, ML and Generative AIKey projects in AI, ML and Generative AI
Key projects in AI, ML and Generative AI
 
“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...
“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...
“Efficient Many-function Video ML at the Edge,” a Presentation from Cisco Sys...
 
ML Platform Q1 Meetup: Airbnb's End-to-End Machine Learning Infrastructure
ML Platform Q1 Meetup: Airbnb's End-to-End Machine Learning InfrastructureML Platform Q1 Meetup: Airbnb's End-to-End Machine Learning Infrastructure
ML Platform Q1 Meetup: Airbnb's End-to-End Machine Learning Infrastructure
 
Slides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdf
Slides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdfSlides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdf
Slides-Артем Коваль-Cloud-Native MLOps Framework - DataFest 2021.pdf
 
Machine Learning Infrastructure
Machine Learning InfrastructureMachine Learning Infrastructure
Machine Learning Infrastructure
 
Cutting Edge Computer Vision for Everyone
Cutting Edge Computer Vision for EveryoneCutting Edge Computer Vision for Everyone
Cutting Edge Computer Vision for Everyone
 
Why is dev ops for machine learning so different
Why is dev ops for machine learning so differentWhy is dev ops for machine learning so different
Why is dev ops for machine learning so different
 
Why is dev ops for machine learning so different - dataxdays
Why is dev ops for machine learning so different  - dataxdaysWhy is dev ops for machine learning so different  - dataxdays
Why is dev ops for machine learning so different - dataxdays
 
Real-Time Voice Actuation
Real-Time Voice ActuationReal-Time Voice Actuation
Real-Time Voice Actuation
 

More from MIDIH_EU

Gare du MIDIH the EC focus on the DIHs network, eDIHs in Digital Europe Prog...
Gare du MIDIH  the EC focus on the DIHs network, eDIHs in Digital Europe Prog...Gare du MIDIH  the EC focus on the DIHs network, eDIHs in Digital Europe Prog...
Gare du MIDIH the EC focus on the DIHs network, eDIHs in Digital Europe Prog...
MIDIH_EU
 

More from MIDIH_EU (20)

GARE du MIDIH the DIHIWARE collaboration platform for mastering your digita...
GARE du MIDIH   the DIHIWARE collaboration platform for mastering your digita...GARE du MIDIH   the DIHIWARE collaboration platform for mastering your digita...
GARE du MIDIH the DIHIWARE collaboration platform for mastering your digita...
 
GARE du MIDIH Open Digital Platforms the adoption of a standards-based open...
GARE du MIDIH   Open Digital Platforms the adoption of a standards-based open...GARE du MIDIH   Open Digital Platforms the adoption of a standards-based open...
GARE du MIDIH Open Digital Platforms the adoption of a standards-based open...
 
GARE du MIDIH MIDIH, towards a flexible, modular and open source reference ...
GARE du MIDIH   MIDIH, towards a flexible, modular and open source reference ...GARE du MIDIH   MIDIH, towards a flexible, modular and open source reference ...
GARE du MIDIH MIDIH, towards a flexible, modular and open source reference ...
 
GARE du MIDIH Digital Manufacturing Platforms in H2020 and in future Digita...
GARE du MIDIH   Digital Manufacturing Platforms in H2020 and in future Digita...GARE du MIDIH   Digital Manufacturing Platforms in H2020 and in future Digita...
GARE du MIDIH Digital Manufacturing Platforms in H2020 and in future Digita...
 
GARE du MIDIH DIH collaboration model
GARE du MIDIH   DIH collaboration modelGARE du MIDIH   DIH collaboration model
GARE du MIDIH DIH collaboration model
 
GARE du MIDIH Methods and Tools to enhance DIHs Digital Transformation powe...
GARE du MIDIH   Methods and Tools to enhance DIHs Digital Transformation powe...GARE du MIDIH   Methods and Tools to enhance DIHs Digital Transformation powe...
GARE du MIDIH Methods and Tools to enhance DIHs Digital Transformation powe...
 
Gare du MIDIH the EC focus on the DIHs network, eDIHs in Digital Europe Prog...
Gare du MIDIH  the EC focus on the DIHs network, eDIHs in Digital Europe Prog...Gare du MIDIH  the EC focus on the DIHs network, eDIHs in Digital Europe Prog...
Gare du MIDIH the EC focus on the DIHs network, eDIHs in Digital Europe Prog...
 
Gare du MIDIH MIDIH general overview
Gare du MIDIH   MIDIH general overviewGare du MIDIH   MIDIH general overview
Gare du MIDIH MIDIH general overview
 
Cemtec lcm midih-presentation-oc2
Cemtec lcm midih-presentation-oc2Cemtec lcm midih-presentation-oc2
Cemtec lcm midih-presentation-oc2
 
PGplant midih-presentation oc2
PGplant midih-presentation oc2PGplant midih-presentation oc2
PGplant midih-presentation oc2
 
Ii3DS novitech-midih_presentation_oc2
Ii3DS novitech-midih_presentation_oc2Ii3DS novitech-midih_presentation_oc2
Ii3DS novitech-midih_presentation_oc2
 
Best route beck et al-midih presentation oc2.
Best route beck et al-midih presentation oc2.Best route beck et al-midih presentation oc2.
Best route beck et al-midih presentation oc2.
 
Smart poly ipf midih-presentation oc2
Smart poly ipf midih-presentation oc2Smart poly ipf midih-presentation oc2
Smart poly ipf midih-presentation oc2
 
Dream bot tractonomy midih presentation oc2
Dream bot tractonomy midih presentation oc2Dream bot tractonomy midih presentation oc2
Dream bot tractonomy midih presentation oc2
 
Proof energy@work midih oc2-demo_day
Proof energy@work midih oc2-demo_dayProof energy@work midih oc2-demo_day
Proof energy@work midih oc2-demo_day
 
DEMOKRITOS - Supreemo midih presentation-oc2_demo_day
DEMOKRITOS -  Supreemo midih presentation-oc2_demo_dayDEMOKRITOS -  Supreemo midih presentation-oc2_demo_day
DEMOKRITOS - Supreemo midih presentation-oc2_demo_day
 
AllbeSmart - E robotic midih-presentation-oc2_demo_day
AllbeSmart - E robotic midih-presentation-oc2_demo_dayAllbeSmart - E robotic midih-presentation-oc2_demo_day
AllbeSmart - E robotic midih-presentation-oc2_demo_day
 
Ah+4diac midih webinar 2020 07-08 reference slide deck
Ah+4diac midih webinar 2020 07-08 reference slide deckAh+4diac midih webinar 2020 07-08 reference slide deck
Ah+4diac midih webinar 2020 07-08 reference slide deck
 
2019 10-23 DIH Campania
2019 10-23 DIH Campania2019 10-23 DIH Campania
2019 10-23 DIH Campania
 
2019 04-30 MIDIH for CONFAPI
2019 04-30 MIDIH for CONFAPI 2019 04-30 MIDIH for CONFAPI
2019 04-30 MIDIH for CONFAPI
 

Recently uploaded

527598851-ppc-due-to-various-govt-policies.pdf
527598851-ppc-due-to-various-govt-policies.pdf527598851-ppc-due-to-various-govt-policies.pdf
527598851-ppc-due-to-various-govt-policies.pdf
rajpreetkaur75080
 

Recently uploaded (14)

Getting started with Amazon Bedrock Studio and Control Tower
Getting started with Amazon Bedrock Studio and Control TowerGetting started with Amazon Bedrock Studio and Control Tower
Getting started with Amazon Bedrock Studio and Control Tower
 
Hi-Tech Industry 2024-25 Prospective.pptx
Hi-Tech Industry 2024-25 Prospective.pptxHi-Tech Industry 2024-25 Prospective.pptx
Hi-Tech Industry 2024-25 Prospective.pptx
 
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
 
05232024 Joint Meeting - Community Networking
05232024 Joint Meeting - Community Networking05232024 Joint Meeting - Community Networking
05232024 Joint Meeting - Community Networking
 
Bitcoin Lightning wallet and tic-tac-toe game XOXO
Bitcoin Lightning wallet and tic-tac-toe game XOXOBitcoin Lightning wallet and tic-tac-toe game XOXO
Bitcoin Lightning wallet and tic-tac-toe game XOXO
 
527598851-ppc-due-to-various-govt-policies.pdf
527598851-ppc-due-to-various-govt-policies.pdf527598851-ppc-due-to-various-govt-policies.pdf
527598851-ppc-due-to-various-govt-policies.pdf
 
123445566544333222333444dxcvbcvcvharsh.pptx
123445566544333222333444dxcvbcvcvharsh.pptx123445566544333222333444dxcvbcvcvharsh.pptx
123445566544333222333444dxcvbcvcvharsh.pptx
 
The Canoga Gardens Development Project. PDF
The Canoga Gardens Development Project. PDFThe Canoga Gardens Development Project. PDF
The Canoga Gardens Development Project. PDF
 
Pollinator Ambassador Earth Steward Day Presentation 2024-05-22
Pollinator Ambassador Earth Steward Day Presentation 2024-05-22Pollinator Ambassador Earth Steward Day Presentation 2024-05-22
Pollinator Ambassador Earth Steward Day Presentation 2024-05-22
 
Acorn Recovery: Restore IT infra within minutes
Acorn Recovery: Restore IT infra within minutesAcorn Recovery: Restore IT infra within minutes
Acorn Recovery: Restore IT infra within minutes
 
Writing Sample 2 -Bridging the Divide: Enhancing Public Engagement in Urban D...
Writing Sample 2 -Bridging the Divide: Enhancing Public Engagement in Urban D...Writing Sample 2 -Bridging the Divide: Enhancing Public Engagement in Urban D...
Writing Sample 2 -Bridging the Divide: Enhancing Public Engagement in Urban D...
 
Sharpen existing tools or get a new toolbox? Contemporary cluster initiatives...
Sharpen existing tools or get a new toolbox? Contemporary cluster initiatives...Sharpen existing tools or get a new toolbox? Contemporary cluster initiatives...
Sharpen existing tools or get a new toolbox? Contemporary cluster initiatives...
 
Eureka, I found it! - Special Libraries Association 2021 Presentation
Eureka, I found it! - Special Libraries Association 2021 PresentationEureka, I found it! - Special Libraries Association 2021 Presentation
Eureka, I found it! - Special Libraries Association 2021 Presentation
 
0x01 - Newton's Third Law: Static vs. Dynamic Abusers
0x01 - Newton's Third Law:  Static vs. Dynamic Abusers0x01 - Newton's Third Law:  Static vs. Dynamic Abusers
0x01 - Newton's Third Law: Static vs. Dynamic Abusers
 

Mo.Le nec-midih_presentation_oc2

  • 1. Speaker: Mauricio Fadel Argerich NEC Laboratories Europe GmbH Model Learning for Cloud-Edge Digital Twin
  • 2. MoLe: Model Learning for Cloud-Edge Digital Twin • With MoLe we aim to simplify the implementation and execution of Digital Twins in digital factories by: • Utilizing FIWARE technologies (NGSI-LD, Scorpio, FogFlow) to dynamically orchestrate a setup to generate a Digital Twin from data and execute it on Edge and Cloud • Reducing effort needed to develop prediction and simulation models by using Knowledge Infusion
  • 3. I4.0Lab – Manufacturing & Assembly process • We have developed our solution for the MIDIH Didactic Factory in Milan • The factory implements the Manufacturing and Assembly process of PCBs • 7 steps, carried out by different stations
  • 4. Translator and Scorpio • Stations in didactic factory use NGSIv2, Scorpio uses NGSI-LD • Translator reads sensor data from the Factory Information Bus and uses Kafka information (topic, key, message schema) to transform it into NGSI-LD entities • Data in NGSI-LD is received by the Scorpio Broker, where it is stored and sent to any subscribers • Scorpio optimizations: • We optimized data handling and serialization, as well as error handling • We optimized Scorpio’s vertical scaling and internal data handling Translator DB Subscriber Subscriber Subscriber Subscriber Scorpio
  • 5. Results: Translator and Scorpio Translator • Worst case delay of 1ms between a message arrival in the Kafka bus and the retrieval in the Python library • Below KPI of 2Hz (sensors sampling frequency) Scorpio • We made a comparison with the updated version of the FIWARE GE Orion called Orion-LD • Scorpio achieved in average half the latency of Orion Scorpio Orion
  • 6. Digital Twin Models • We implemented Digital Twins (DTs) for the different stations in the M&A process • DTs are programmable objects that can be instantiated for real-time monitoring and simulations • DTs of Front Cover Magazine and the Press Station • These DTs implement specific models to: • detect the current status of the station based on its current sensor and actuators data • predict energy usage based on the same data
  • 7. Knowledge Infusion • Energy usage prediction à pure ML • State inference model à Knowledge Infusion (KI) = ML + domain knowledge • Domain knowledge is infused through Knowledge Functions (KFs) • KFs output a single value: a label. • Functions implement human provided logic and utilize facts derived from internal and external knowledge bases • Types of knowledge functions: Weak and Strong • KI creates a Knowledge Model that serves two purposes: • Data augmentation: improves data quality by creating new features or labelling data • Robustness: allows us to correct some obviously wrong outputs of the ML model during runtime ML Model Knowledge Model KI Model
  • 8. FogFlow • FogFlow was extended to serve ML models as serverless fog functions • ML models are implemented based on FogFlow ML operators • To implement a FogFlow ML model, we follow 3 steps: 1. Model registration: to register a ML through a web-based GUI 2. Model deployment: to create and deploy a serverless fog function in FogFlow to run the ML model 3. Model serving: to apply the ML model inside the deployed function instances to produce the detection/prediction result of the input data and then update the state of the corresponding DT
  • 9. Results: Digital Twin Models • Model for the station’s state • Small high quality dataset: manually labeled 10m of data (2Hz, 1200 data points total) as “Idle” or “Working” • Larger noisy dataset: around 50m or 6000 data points, labelled with 2 simple functions • We used the small dataset to train a Random Forest Classifier (RFC) using all the features (45) • First 600 samples to train RFC and last 600 to evaluate it • The RFC achieved an accuracy of 82.11% on its test set • We also implemented a Knowledge Model (KM), based on two programmable functions • Each function took no more than 5 minutes, they check variable values and return state of the machine • Test accuracy is 70% Performance of RFC on test set when trained with 50% of manually labeled data Performance of KM on test set with 2 simple labelling functions S L train test à
  • 10. Results: KI for Digital Twin Models • We can utilize KM to label larger unlabeled dataset and train a ML model with it • Hopefully, the ML will learn to filter out the noise • The ML model is trained with 10x more data as before • We re-trained the RFC with these larger noisy dataset • Slight improvement in its performance: 82.33% • Without any manually and costly labeled data! • KM took us about 10 minutes to implement • We have also implemented a KI model that utilizes a supervision function • This function verifies the value of certain variables and forces the output of the ML model • This function represents a layer of safety • Accuracy in our tests remained the same
  • 11. Results: Energy Consumption Prediction • We implemented a Random Forest Regressor to deal with correlated and non- informative variables • Energy consumption is influenced by activity of the station, logged by its sensors • There are variables which are correlated and others are nearly constant throughout the activity, it seems we might have partial visibility • We trained it using 80% of the full time series data for Press Station and kept the 20% remaining as test set • The Random Forest Regressor obtained a Mean Absolute Error of 2.75. train test à Results on test set
  • 12. KPIs KPI 1: Data velocity 1000 msgs per second >> 2 msgs per second (sampling frequency of sensors in digital factory) KPI 2: Generation of knowledge graph for the DT The Translator is capable of generating a graph structure in the form of NGSI-LD Entities. This is done automatically using provided meta data from the factory message bus. KPI 3: Accuracy of KI for DT Accuracy of ML model trained with hand labeled data: 82.11% Accuracy of Knowledge Model: 70.50% Accuracy of ML model trained with data labeled by Knowledge Model: 82.33% KPI 4: Accuracy of DT Refinement (Strong KF) Accuracy of KI model with strong KF: 82.33% Note: Same accuracy as without refinement because Strong KF did not find any necessary correction
  • 13. Lessons learnt NGSI Translator: Valuable tool for extracting knowledge from raw data, allows for more flexibility (NGSIv2, NGSI-LD). You can find the translator at https://github.com/ScorpioBroker/ScorpioBroker/tree/feature-82/NGSILDTools/NGSILDTranslator KAFKA NGSI-LD Integration: Kafka is a good choice as it is wide spread and has excellent performance. Recommendation: use key on Kafka messages so an identifier is attached to the data. Digital Factories Data: Data heterogeneity between factories is still very high. Opportunity: tools/techniques to join data from different factories are valuable and needed! Recommendation: it’s beneficial to publish example data from MIDIH factories. Knowledge Infusion: it enabled us to train a classifier with no manually labeled data, achieving high accuracy. KI shows great potential to reduce effort of creating ML models. KI, ML models’ performance and execution in FogFlow: Models achieved good accuracy but we believe this can be further improved.