Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
APPLICATION
DEVICES
Model-Based Design
with MATLAB and Simulink
to shorten the design of a new
infusion pump
Fresenius Approach
Claude VITTORIA, Fresenius Kabi
Model-Based Design to shorten the design of a new infusion pump
About Fresenius Kabi
Medical market issues to guide Fresenius decisions
Model-Based Design in the product development
Lessons learned and perspectives
MATLAB EXPO – October 2nd 2014 © Copyright Fresenius Kabi AG 2
Model-Based Design to shorten the design of a new infusion pump
About Fresenius Kabi
Medical market issues to guide Fresenius decisions
Model-Based Design in the product development
Lessons learned and perspectives
© Copyright Fresenius Kabi AG 3MATLAB EXPO – October 2nd 2014
Fresenius organization
© Copyright Fresenius Kabi AG 4
• Infusion devices• Dialysis• Equipments• Services
• Employees: 86 000
• Infusion therapies• Nutrition therapies• Medical Devices• Transfusion
• Employees: 30 000
• Hospital Operations
• Employees: 37 000
• Engineering & Services• Healthcare facilities
• Employees: 4400
• Sites in over 100 countries• Turnover: 19 200 M€• Employees: 170 000
MATLAB EXPO – October 2nd 2014
FRESENIUS KABI AG Medical Devices : A global solution
© Copyright Fresenius Kabi AG 5MATLAB EXPO – October 2nd 2014
FRESENIUS KABI AG Medical Devices : A global solution
© Copyright Fresenius Kabi AG 6MATLAB EXPO – October 2nd 2014
FRESENIUS KABI AG Medical Devices : A global solution
© Copyright Fresenius Kabi AG 7MATLAB EXPO – October 2nd 2014
FRESENIUS KABI AG Medical Devices : A global solution
© Copyright Fresenius Kabi AG 8MATLAB EXPO – October 2nd 2014
Model-Based Design to shorten the design of a new infusion pump
About Fresenius Kabi
Medical market issues to guide Fresenius decisions
Model-Based Design in the product development
Lessons learned and perspectives
© Copyright Fresenius Kabi AG 10MATLAB EXPO – October 2nd 2014
Stakes & ChallengesWhy Model Based Design ?
Improve design process
– Scenario validation
– Increase collaboration
– Knowledge sustainability
– Code validation for verification
Reduce time to market
– Speed up algorithm validation
– Automatic code generation
Prepare Medical Certification
– Refine requirements & traceability
– Model verification & validation
11© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
How to get started ?
– Blank page syndrome
– Which Process and Methodology ?
Which tools to achieve our objectives ?
– Combination of physical model and control algorithm
– Simulation, rapid prototyping and production code generation
– Certification
Mathworks solution
– Tools (Simulink, Simscape, Stateflow)
– Services (training, follow-up)
– Proven in use in other industries
Stakes & ChallengesSimulink to address MBD for Medical devices
12© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Simulink or Simscape ?
Why Simscape ?
– knows I.S. units, simulation faster, rigor
– Simulink assists to define and to tune the equations
Physical Model
13© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
)()()()( tFtkxtxbtxm
Simulink or Stateflow ?
– No continuous regulation (PI, PID) in our products
Why Stateflow ?
– FSM theory (Moore, Mealy, actions, events)
– Lifecycle representation
– Debug tools
– Easy time segregation for code generation
Control algorithms
14© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Model-Based Design to shorten the design of a new infusion pump
About Fresenius Kabi
Medical market issues to guide Fresenius decisions
Model-Based Design at Fresenius
Lessons learned and perspectives
15© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Model-Based Design at FreseniusOverview
16
System requirements
Modelrefinement
Processor In the Loop
On-target prototyping
Hardware In the Loop
Rapid control prototyping
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
New product shall be compliant with
Medical Devices 60601-1 3th edition
and60601-2-24 2th edition
Make a simple physic
– Stay dynamic, let’s use parameters
– Simulate and validate the model without physical constraints
Make the first iteration of the algorithms
– Validate hypothesis and refine the requirements
Model-Based Design at FreseniusModeling
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Processor In the Loop
On-target prototyping
Hardware In the Loop
Rapid control prototyping
System requirements
Modelrefinement
MATLAB
Simulink
Stateflow
Simscape
MATLAB Coder
Simulink Coder
Simulink Verification & Validation
MATLAB Report Generator
Simulink Report Generator
17
04/2013 Licenses purchase
06/2013First Simulink
models
12/2013First Simulink
models
Simulation
– Try...Mine...Try again
– Quickly & Faster
– Improve targeted performance
Cost reducing conception
– Validate requirements
– Track non reasonable requirements
– Refine requirements to help the choice of technologic solutions
Model-Based Design at FreseniusHypothesis & simulation
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Processor In the Loop
On-target prototyping
Hardware In the Loop
Rapid control prototyping
System requirements
Modelrefinement
18
MATLAB
Simulink
Stateflow
Simscape
MATLAB Coder
Simulink Coder
Simulink Verification & Validation
MATLAB Report Generator
Simulink Report Generator
04/2013 Licenses purchase
06/2013First Simulink
models
11/2013Validate
requirements
12/2013First Simulink
models
02/2014Validate
requirements
Use your conception and your algorithms
– Stop to modify existing to get a mechanic animated
– Break the existing limitations
– Validate technologic solutions
– Confirm model and simulation
Develop validation plan earlier
– Better maturity of plan and team
Model-Based Design at FreseniusRapid prototyping
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Processor In the Loop
On-target prototyping
Hardware In the Loop
Rapid control prototyping
System requirements
Modelrefinement
MATLAB
Simulink
Stateflow
Simscape
MATLAB Coder
Simulink Coder
Simulink Verification & Validation
MATLAB Report Generator
Simulink Report Generator
19
04/2013 Licenses purchase
06/2013First Simulink
models
08/2013Rapid
Prototyping
11/2013Validate
requirements
12/2013First Simulink
models
02/2014Validate
requirements
04/2014Rapid
Prototyping
Validate algorithms and their precisions
– Keep advance, choose a compatible target (Texas Instruments toolbox)
– Code coverage tests
– Valid calibration process
Create functional tests for the product integration
– HW filters fit to the mechanical behavior
– Push the limits of HW
Model-Based Design at FreseniusProcessor-in the-Loop and on-target
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Processor In the Loop
On-target prototyping
Hardware In the Loop
Rapid control prototyping
System requirements
Modelrefinement
MATLAB
Simulink
Stateflow
Simscape
MATLAB Coder
Simulink Coder
Simulink Verification & Validation
MATLAB Report Generator
Simulink Report Generator
20
04/2013 Licenses purchase
06/2013First Simulink
models
07/2014Tests on
HW
08/2013Rapid
Prototyping
11/2013Validate
requirements
12/2013First Simulink
models
02/2014Validate
requirements
04/2014Rapid
Prototyping
05/2014Tests on
HW
Repeatability of the validation plan
– No regression, play the same tests again, again and again
Complex tests
– Hypothetic failures, unlikely use
– Push the limits of Sw/Hw
Model-Based Design at FreseniusHIL based on Simscape model
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Processor In the Loop
On-target prototyping
Hardware In the Loop
Rapid control prototyping
System requirements
Modelrefinement
MATLAB
Simulink
Stateflow
Simscape
MATLAB Coder
Simulink Coder
Simulink Verification & Validation
MATLAB Report Generator
Simulink Report Generator
21
04/2013 Licenses purchase
06/2013First Simulink
models
07/2014Tests on
HW
08/2013Rapid
Prototyping
11/2013Validate
requirements
12/2013First Simulink
models
02/2014Validate
requirements
04/2014Rapid
Prototyping
05/2014Tests on
HW
Document in the model
– Requirements and justifications
Verification & Validation
– Referenced requirements from external documents in the model
– Create tests and reference for validation plan
– Create automatic regression test suits
Report generator
– Generate conform documents to the quality assurance
Model-Based Design at FreseniusRequirements & Traceability
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Processor In the Loop
On-target prototyping
Hardware In the Loop
Rapid control prototyping
System requirements
Modelrefinement
MATLAB
Simulink
Stateflow
Simscape
MATLAB Coder
Simulink Coder
Simulink Verification & Validation
MATLAB Report Generator
Simulink Report Generator
22
04/2013 Licenses purchase
06/2013First Simulink
models
07/2014Tests on
HW
06/2014Generate
documentation
08/2013Rapid
Prototyping
11/2013Validate
requirements
12/2013First Simulink
models
02/2014Validate
requirements
04/2014Rapid
Prototyping
05/2014Tests on
HW
06/2014Generate
documentation
Model-Based Design to shorten the design of a new infusion pump
About Fresenius Kabi
Medical market issues to guide Fresenius decisions
Model-Based Design in the product development
Lessons learned and perspectives
23© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Achieving technology together
– How to integrate in the existing ?
– Which perimeter ?
Try to meet any type of interrogations
– Theory vs Practice
Take care to be compatible with our solution choices
– Technical and tooling
Tools are skill-less but helpful
– Find the Know-how
Warned about
24© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
BenefitsExplore technical solutions
25© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Validate technical choices early
– Simulation
– Rapid prototyping
Get a specification faster with a better confidence
– Explore more solutions through simulation
– Justify choices
BenefitsTeam work
26© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Provide solutions to help the teams
– Rapid prototyping
– SIL/PIL
– Code generation
Greater interaction between the teams
– Model
– Prototypes
– Tools
– Specifications
MATLAB EXPO – October 2nd 2014 © Copyright Fresenius Kabi AG 27
BenefitsTests & Validation
Think tests and validation earlier
– Document
– Trace requirements
– Create tests
Quality of tests
– Exhaustively
– Define the exact behavior of the product
Time and effort
– Better vision
– Better control
– Good practice in place
Next step
© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014 28
Run more tests on Simulink models
– Formal proof
Production code generation
– Embed algorithm into microcontroller (volumetric pump and syringe pump)
Develop new ranges of pumps
– Reusability
Model-Based Design to shorten the design of a new infusion pump
29© Copyright Fresenius Kabi AGMATLAB EXPO – October 2nd 2014
Thanks for your attention