Predictive Maintenance (with a Thermal Imaging Camera, vibration sensors, etc.)

need to fill out these templates: Documentation Sub-committee


PTL- Vladimir Suvorov - 17 September 2020 through 17 September 2021

Use Case Details:

<add in the taxonomy image from the white paper>


Use Case  -  Predictive Maintenance using a Thermal Imaging Camera


Attributes

Description

Informational

Type

New

New

Industry Sector

IoT Device Edge


Business driver

Predictive Maintenance


Business use cases

Many devices give off hints that they will need to have maintenance earlier than their scheduled maintenance.  Through Machine Learning (ML), we can create models that will allow us to know that a device will soon need maintenance.  For many machines, we can gain a great deal of information on the health of the device by looking at the temperature of the device.  This requires collecting the data and then sending it to a Historian or similar device.  These data points can be sent to the cloud to be modeled.

Other requirements

  • Need to take the current temperature of the device and react in near real-time to rising temperature
    • Example: If over 150 C- send out a warning to an email list, show warning on a UI
      if over 180 C trigger light or horn
      if over 200 C trigger shutdown process

Other variations:

Monitoring restricted spaces

  • If a human enters in a space, 
    • first level of restriction- sound an alarm and turn on lights
    • second level- start the shutdown process

Predictive maintenance: There are many different types of models.  For example, many models do not need to be done in real-time.  Thus, the data can be sent to the Cloud and processed.  The data is not time critical, so if there is a delay in sending/receiving data, the data will need to be stored and then sent when the network is available. 


Yet, there are many scenarios, where real-time or near real-time is required.  An example of this would be a machine reaching a maximum temperature.  As it approaches this, we would want to send out a warning and then if it reached this critical temperature, the device needs to be shut down.


For this type of scenario, there needs to be a server or space on the IoT gateway that can process the data in real-time. 


Business Cost - Initial Build Cost Target Objective

Cost is only for the hardware- 


Business Cost – Target Operational Objective

varies widely depending on accessories.  The IoT Gateway can be under $500 to over $5,000


Security need

Because of the remoteness of the devices, need the ability to control ports (turn on/off)


Regulations

Varies depending on local regulations


Other restrictions



Additional details




Family- IoT Device Edge-

Use Case Attributes

Description

Informational

Type

New


Blueprint Family - Proposed Name

IoT-Device Edge

There are many possible UCs that would be IIoT, so these only are designed to handle Predictive Maintenance UCs 

Use Case

Predictive Maintenance using a FLIR Camera

See below

Blueprint proposed

Predictive Maintenance- Using FLIR Camera



Initial POD Cost (capex)

Varies widely depending on the Blueprint


Scale of Servers

one at the User Edge

this is the IoT Gateway

Applications (Edge Virtual Network Functions)

EVE


Power Restrictions

None/Varies

  • none for the FLIR, but another blueprint might need it

Preferred Infrastructure orchestration

Docker/K8 - Container Orchestration

OS - Linux


Additional Details



BluePrint (Species) - Predictive Maintenance- with a Thermal Imaging Camera

Case Attributes

Description

Informational

Type

New


Blueprint Family - Proposed   Name

IoT Device Edge

IIoT == Industrial Internet of Things

PM == Predictive Maintenance 

Use Case

Any Predictive Maintenance UC that is on the shop floor

With a few modifications, it is possible to change this blueprint to meet many similar Use Cases

Blueprint proposed Name

Predictive Maintenance using a FLIR Camera


Initial POD Cost (capex)

Under $20k

FLIR Camera-

IoT Gateway- Advantech- Model UNO

LFEdge's Adam or similar

Fledge 

This is the set up for the FLIR Fledge/EVE demo

  • the demo uses ZEDEDA instead of LF Edge's Adam

Scale & Type of Server

1 IoT Gateway, a server on the edge is not needed 

This is on the customer edge, thus there is no server.  The IoT Gateway will handle the connection to the internet.

Applications

Fledge, Ubuntu, code for the demo


Power Restrictions

NA

none of the devices require power that is outside of a normal wall socket

Infrastructure orchestration

EVE

VM- Ubuntu

EVE acts as the OS and will have a containerized version of Ubuntu and Fledge on it 

SDN (Software Defined Networking)

None


Workload Type

  • Containers (Tensoflow, Keras containers)
  • VM- Ubuntu


Additional Details




PTL- Vladimir Suvorov - 17 September 2020 through 17 September 2021



Committer

Committer

Company

 Committer Contact Info

Committer Bio

Committer Picture

Self Nominate for PTL (Y/N)

@bill hunt

Dianomic

bill@dianomic.com




Shiv Ramamurthi

ArmShiv.Ramamurthi@arm.com


Cplus Shen

Advantech

Cplus.Shen@advantech.com.tw




Ashwin GopalakrishnanDianomicashwin@dianomic.com


Erik NordmarkZededaerik@zededa.com


Daniel LazaroOSIsoftdlazaro@osisoft.com


Aaron WilliamsIndividualaaron@wi5s.com


Vladimir SuvorovAi Solutionshello.fleandr@gmail.com

Y


Contributors: 

Tina Tsou (Deactivated)