Embed Size (px)
Citation preview
Model Based Design @ Getinge in Solna
Who am I?Magnus Nord - Model Based Design Enthusiast
8 May 2019
MBD at Getinge in Solna
Page 2
Presentation Road Map
8 May 2019
Page 3
On a personal noteA fairly normal presentation
Getinge GroupSolna – a part of Acute Care Therapies
8 May 2019
Page 4
Advanced Monitoring
Products
8 May 2019
Page 5
AnesthesiaVentilation
Solna
Numbers
• Tot Solna ~ 450
• R & D: ~ 150
• Matlab/Simulink: ~ 20 licences
• MBD ~ 4-8
• Revenue: ~ 2000 MSEK
Products
8 May 2019
…and much more
Page 6
Outside Solna
Operating RoomSterilization
Model Based Design at GetingeSolna
8 May 2019
Page 7
Model Based Design
8 May 2019
Page 8
Maturity Assessment
Control Design
dSpace
Hand Written
Code
Model Based Design
8 May 2019
Page 9
A Timeline
Time
MBD efficiencyAnesthesia
Machine
Model #2
Control Design
Generated CodeControl Design
Generated Code
Patient Model
Anesthesia
Machine
Model
Stateflow
Ventilator Model
Control Design
Generated Code
PIL
Rapid Prototyping
1994 2010 2015 2020
Model Based Design
8 May 2019
Page 10
ResultsResults
• Overall good performance and development speed!
• From prototype to production code in short time.
• Several ventilation and anesthesia products and
ventilation modes with generated code.
• Promising results from Lo-fi model and new control
strategy. Model will increase development speed of
future control improvements.
• Incremental implementation. Let MBD-grow into place.
Model Based Design
8 May 2019
Page 11
A Timeline
Time
MBD efficiencyAnesthesia
Machine
Model #2
Control Design
Generated CodeControl Design
Generated Code
Patient Model
Anesthesia
Machine
Model
Stateflow
Ventilator Model
Control Design
Generated Code
PIL
Rapid Prototyping
1994 2010 2015 2020
Control Design
dSpace
Hand Written
Code
Ventilator
Products - Solna
8 May 2019
Page 12
Products from an automatic control perspective
Pressure Control Flow Control
Control Logic
Anesthesia Machine
Pressure Control Flow Control
Control Logic Anesthesia Control
Model Based Design
Anesthesia
Machine
Model
8 May 2019
Page 13
Model Components and challenges
Patient
Gas
Modules
Reflector Absorber
Circle
Breathing
System
Model Based Design
8 May 2019
Anesthesia
Machine
Model
Page 14
Anesthesia Model
Model Based Design
Anesthesia
Machine
Model
8 May 2019
Page 15
Model Components
Gas
Modules
Reflector Absorber
Challenges
• Minimize Gas Consumption
• Diffusion – bidirectional flow
• Reflector
• Absorber
• Nozzle characteristics in Gas
Module
• Model Speed!!
• Non minimum phase control
with long time-varying group
delay.
Patient
Circle
Breathing
System
Model Based Design
Anesthesia
Machine
Model
8 May 2019
Page 16
Incremental improvements
Machine
+
Patient
Model
High Fidelity
Model
3x Speed
Save
Environment
Low Fidelity
Model
10x Speed
Accuracy
where
needed
Model Based Design
8 May 2019
Page 17
Stateflow
Stateflow
Advantages
• Readability
• “Code looks like a
design document”
• Intuitive coding
• Surrounding controller logic
sometimes more complex
than the controllers –
Stateflow helps!
Ventilator Logic
Ventilation Mode
Working well for
• Alarm handling
• Metric Calculations
• Ventilator Logic
Rapid Prototyping
Model Based Design
8 May 2019
Page 18
Rapid Prototyping and Data Acquisition
High Level Design
Simulink
Target
dSpace
External
Mode
Wafer
Thin
Wrapper
Low Level Debugging
Atolic
Eclips Based
SDKSWV
debugging
CubeMX
Manual
reduction:
4500 pages ->
350 pages
On a personal noteEnlarge the Simulink ecosystem
Technical Paradigm Shifts and Division of Labor
Everyday Tools
8 May 2019
Page 19
Model Based Design
8 May 2019
Page 20
The future in high level design – leverage on good quality tools. Expand Simulink Ecosystem
Model Based Design
8 May 2019
Rapid Prototyping
Page 21
The future in high level design – leverage on good quality tools. Expand Simulink Ecosystem
High Level Design
Simulink
Hadoop
AWS
xCode
Qt
Xilinx – Zynq
CubeMX
Model Based Design
Anesthesia
Machine
Model
8 May 2019
Page 22
Everyday tools that work
Machine
+
Patient
Model
High Fidelity
Model
3x Speed
Save
Environment
Low Fidelity
Model
10x Speed
Accuracy
where
needed
Obstacles on the MBD road…. and ways to remove themPDF version
8 May 2019
Page 23
Model Based Design
8 May 2019
Page 24
MBD-hurdles and how to jump over them
MBD has a proven track record of
reducing time and cost for
complex development projects.
Where do we need improvements?
What hurdles are blocking the MBD-runner?
MBD
Model Based Design – function development
8 May 2019
Page 25
MBD-hurdles and how to jump over them
Solution #1
areas
• Alarms
• Metrics
• Control Logic
• Ventilator
alternative
Legacy Code
BlockageSolution #1
• Incremental Improvements
• Thesis Projects
• Replacement strategy
• Add “structural” improvements to
each project
Solution #2
• Plant model for
legacy code
• Interface to legacy
code
Model Based Design
8 May 2019
Page 26
MBD-hurdles and how to jump over them
Collaboration
BlockerProblem
• Merge Tool
Solution #1
• Small team helps. Its easier to
collaborate in small teams. Model and
file ownership.
Solution #2
• Merge Tool improvements are needed!
We need from
Model Based Design
8 May 2019
Page 27
MBD-hurdles and how to jump over them
Argument: Initial
Cost of Model
Based DesignSolution
• Show success stories
• Bombardier
• Scania
• Practically all car
manufacturers
• Getinge
• etc.....
Solution
• Show hidden costs without MBD
• A validated model always adds IP -
even if its done in preparation for
next project.
Conclusions
8 May 2019
Page 28
Model Based Design
8 May 2019
Page 29
Take away from this speech
Take Away
• Incremental implementation
• Let MBD grow into place
• Model Based Design Works
• Development speed has been
proven.
• Validated model always adds vital
IP – even if its done in preparation
for next project.
Model Based Design
8 May 2019
Page 30
Take away from this speech
Personal Take Away
• Leverage on good quality tools
• Find tools and processes that enables
job enrichment and implementation
efficiency.
Presentation Road Map
8 May 2019
Page 31
On a personal note
Questions?
A fairly normal presentation
