matlab & simulink in validating future technologies with ... file- s. balakumaran . renault...
TRANSCRIPT
Embedded Software|10 July 2014 Renault Nissan General
MATLAB & Simulink in validating future technologies with Real time test data in Automotive Engineering
- P. Gandhimathi - S. Balakumaran
2 Renault Nissan General |
Challenges in validating new applications
Agenda
This is an example text. Go ahead and replace it
Introduction
2
3
4
New Application
Application Validation
Accelerator mode of simulation and Simstate
5 Simulation approach
6 Tools support
Benefit 7
Alternate method of validation
Test scenario creation
Validation with huge data
1
3 Renault Nissan General |
Introduction
Increasing software applications in Automotive industries.
The usage of software in automotive industry is inevitable and its share keeps on
growing beyond 80%.
The percentage will rise more than 90% within the next decade.
Applications in automotive – Safety critical
Effective testing of the applications from software development phase itself.
Software development in Simulink
MATLAB & Simulink are provided with lot of features for software verification and
validation.
4 Renault Nissan General |
Application validation
General application validation can be with any of the following test inputs,
• Manual test cases based on requirement
• Auto generated test cases
• Real time vehicle input data.
5 Renault Nissan General |
New Application
OEMs focus on adding new features to the vehicle in order
• To satisfy the more strict safety requirements
• To provide better driving experiences to customer
• To compete in Market
Initial phase of any new application would be to simulate it on software level.
This simulation should prove
• Application functionality
• That it will work in real time environment
The challenge would be to prove that the new application will work in real time!
6 Renault Nissan General |
Challenges in validating new applications
Collecting test data for new applications in real time poses various challenges to us such as,
• Long vehicle run with specific safety critical drive conditions.
• Time consumption.
How do we test the new application with real time data with these challenges?
And How MATLAB & Simulink helps us to make it simple?
7 Renault Nissan General |
Alternate method of validation
• The alternative approach
– Collect the real time test data for short distance drive in real time with specific
conditions necessary
– Create complete test cases with combinations of the collected real time test data
– Validate the software (Simulink model) by simulation in MATLAB
8 Renault Nissan General |
Test scenario creation
• Extracting vehicle data
– Set of CAN data from vehicle are taken and converted to .mat file for loading into the
MATLAB workspace.
• Creating scenarios
– Combinations of set of input data are formed in MATLAB with the help of a tool developed
with m-script.
• Model validation
– Validation by simulating the model with input combinations using the tool.
– Result analysis with the plots generated out of simulation against inputs
9 Renault Nissan General |
Validation with huge data
• Simulation process needs more memory because of
– Lengthy test cases created in .mat
– More temporary memory allocation need for complex Simulink models
– Large data storage for result analysis
• Allocated MATLAB heap memory for workspace is not sufficient to handle the simulation with
huge test data.
• Validating with lesser input data will not provide the intended result and requires user to stay
nearby for long time for successive simulations.
10 Renault Nissan General |
Accelerator mode of simulation and Simstate
• With “Normal” mode of simulation, all the internal states and variables are stored for every sample
time – Leading to Out of Memory Error
Normal
mode
11 Renault Nissan General |
• Hence moving to Accelerator mode of simulation – which stores only the final output values
Accelerator mode of simulation and Simstate Contd…
Accelerator
mode
12 Renault Nissan General |
• Need for the final state information & accelerated simulation
• With Simstate settings in Simulink, Final state of a smaller simulation can be saved and restored
as the initial state for the next simulation.
Simulation 1
State 1
State 2
State 100
Simulation 3
State 200
State 201
State 300
Simulation 2
State 100
State 101
State 200
Accelerator mode of simulation and Simstate Contd…
13 Renault Nissan General |
Simulation approach
• This approach helps in clearing and saving of the workspace variables in-between the
simulations, thereby effectively reusing the memory without causing memory error.
•test1.mat,
•test2.mat,
•test3.mat
•etc.
Load Test case(.mat)
•[test1.mat+test2.mat+test3.mat etc.]
Create Test scenario
•test1.mat (load, simulate, save, clear)
•test2.mat (load, simulate, save, clear)
•etc. Simulation
(Accelerator Mode)
•Result.mat
Consolidated Result
•Plot
Result analysis
14 Renault Nissan General |
Tools support
• Software Validation taken closer to real time
Validation Tool
M-script (medium of
development)
GUIDE
(Effective User interface)
Simulation settings
(Handling complex simulation)
Plot (Detailed Result
Analysis)
15 Renault Nissan General |
Benefit
• MATLAB, Simulink helps us to validate our new applications with real time test data in easy way
with less number of resources and effort
16 Renault Nissan General |
Thank you for
your attention!
Any Questions?