1
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
1
USING THE MATLAB TOOLSET TO IMPROVE EFFICIENCY IN THE EOBD CALIBRATION
PROCESS
N. Wright and R. DraneFord Motor Company Limited, UK
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
2
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
2
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
3
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
4
Background
• Reduced development cycle time• Reduced usage of prototype vehicles• Reduced testing costs• Re-usability of test data
Ford has developed a number of CAE tools to improve efficiency in the EOBD calibration process. Benefits of this approach include:
3
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
5
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
6
Historical Calibration ProcessCollect Mapping/Calibration
Data
Post Process Mapping Data and Generate Initial Calibration
Individual task specific programs and spreadsheets
Collect More Data
Run Calibration on Vehicles to Validate
Collect Data for Secondary Calibration Task
IterativeProcess
Post Process Data for Secondary Calibration task
Individual task specific programs and spreadsheets
Collect More Data
Run Total Calibration on Vehicles to Validate
Collect Sign-Off Evidence Data
Key Calibration Parameter Change
OKNot OK
Not OKOK
4
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
7
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
8
New Calibration ProcessCollect Data on Vehicles for Entire
Calibration and Validation Input Signals for Simulink Simulation Tools
Integrated Calibration and Validation Process Using Matlab Toolset:
Custom Data Processor ToolMBCModel
CAGESimulink Simulation Tools
IterativeProcess
Collect Sign-Off Evidence Data
Key Calibration ParameterChange
5
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
9
Matlab Toolset Integrated Calibration Process
Collected Data
Data Processor ToolProcess data collected for
calibration generation
MBC ModelModel calibration Surfaces
CAGEGenerate initial calibrations
SpreadsheetsGenerate initial
calibrations
Simulink Simulation ToolsRun simulation models with all
collected data sets
Tweak Calibration
Data Processor ToolConduct validation analysis on
output from simulink models
Collect Sign-Off Evidence Data
Not OK
OK
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
10
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
6
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
11
Simulink Simulation ToolsCollected Data
Data Processor ToolProcess data collected for
calibration generation
MBC ModelModel calibration Surfaces
CAGEGenerate initial calibrations
SpreadsheetsGenerate initial
calibrations
Simulink Simulation ToolsRun simulation models with all
collected data sets
Tweak Calibration
Data Processor ToolConduct validation analysis on
output from simulink models
Collect Sign-Off Evidence Data
Not OK
OK
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
12
Simulink Simulation Tools
• Desktop calibration and validation• Data batch processing• Auto calibration• System robustness testing• Standard controller allows connection to any
correctly formatted Simulink Model
Ford in association with The MathWorks have developed a number of Simulink Powertrain Control System Strategy Models. Features of the tools:
7
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
13
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
14
Generic Simulink GUI - Tool Format
8
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
15
Generic Simulink GUI - Tool Format
Controller Automatically Recognises Calibration Parameters, Inputs and
Signals and configures GUI
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
16
Generic Simulink GUI - Tool Format
Load Calibration Information and Data
9
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
17
Generic Simulink GUI - Tool Format
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
18
Generic Simulink GUI - Tool Format
Edit Calibration Scalars
Select Input Conditions Using Recorded Data
or Constant
Select Which Inputs and Signals to Display
10
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
19
Generic Simulink GUI - Tool Format
Edit Table Function
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
20
Generic Simulink GUI - Tool Format
View Flags
View Variables
View Selected Point
11
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
21
Generic Simulink GUI - Tool Format
View Data in Standard units or Normalise
Variables
Zoom Option on Display
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
22
Generic Simulink GUI - Tool Format
Activate History Option
12
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
23
Generic Simulink GUI - Tool Format
History Option Allows Viewing Effects of
Calibration Changes
Select Up to 4 History Files
Review Calibration Differences Between Files
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
24
Generic Simulink GUI - Tool Format
Save Options
13
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
25
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
26
HEGO Simulink GUI - Batch Processing
14
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
27
HEGO Simulink GUI - Batch Processing
Select Min & Max Values
Select Number of Steps
Select Calibration Variables to
Sweep
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
28
HEGO Simulink GUI - Batch Processing
15
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
29
HEGO Simulink GUI - Batch Processing
View all Sweeps
Rotate Facility on Graphics
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
30
HEGO Simulink GUI - Batch Processing
Statistics on all Sweeps
Distribution Data can be Viewed in Bar Form
or Line Form
16
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
31
HEGO Simulink GUI – Auto Calibration
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
32
HEGO Simulink GUI – Auto Calibration
Select Best Calibration
View Individual Sweeps
Statistics on Individual
Sweep
17
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
33
HEGO Simulink GUI – System Robustness
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
34
HEGO Simulink GUI – System Robustness
Select Signal Generator
Set up Base Signal Select
Robustness Limits
18
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
35
HEGO Simulink GUI – System Robustness
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
36
HEGO Simulink GUI – System Robustness
Calibration Not Robust Over all Robustness
Limits
19
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
37
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
38
Data Processor ToolCollected Data
Data Processor ToolProcess data collected for
calibration generation
MBC ModelModel calibration Surfaces
CAGEGenerate initial calibrations
SpreadsheetsGenerate initial
calibrations
Simulink Simulation ToolsRun simulation models with all
collected data sets
Tweak Calibration
Data Processor ToolConduct validation analysis on
output from simulink models
Collect Sign-Off Evidence Data
Not OK
OK
20
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
39
Data Processor Tool - Features
• Flexible – has countless applications• Graphical display of imported data• Data filtering to either exclude data or to select data for
export prior to analysis• Has access to Matlab functions• Allows use of users own written functions• Enables batch processing of data from multiple files• Results can be imported into MBC for modeling• Results can be exported in .csv format for spreadsheets• Tasks are saved as templates, similar to MBC.
The data processing tool is used to manipulate and analyse data generated from vehicle or from simulink models. Specific features include:
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
40
Data Processor Tool – Graph View
Combustions
Misfires
Raw Misfire Detection Signal
Generated Misfire Signal(occurs 6 events prior toactual misfire)
21
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
41
Variable Editor Change Plot Properties
Graph View - Options
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
42
Filter Sets
Filtering Tool Buttons
Graph View – Filtering
22
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
43
X filter -- Removes data along the X axis Y filter -- Removes data along the Y axis XY filter -- Removes rectangle area of data
Data can be removed graphically by use of mouse click and drag operation:
Filtering – Removing Data
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
44
Export filters can be used to select data that is requiredto be kept for analysis:
X Export Filter – Selects data along the X axisY Export Filter – Selects data along the Y axisXY Export Filter – Selects data in a rectangular area
Table View:
Filtering – Selecting Data
23
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
45
Data Processor Tool – Table View
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
46
SNRIndex = Mean(comb) – Mean(mis)STD(Comb)
User Defined Function:
Table View – Columns of Results
24
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
47
MBCModel
Spreedsheet
Data Processor Tool – Batch Run
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
48
Simulink GUI Template configured with desired calibration:
ValidationResults Files
Data Processor Tool – Simulink Model Analysis
25
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
49
Agenda
• Introductions and Agenda N. Wright
• Background N. Wright
• Historical Calibration Process R. Drane
• New Calibration Process R. Drane
• Simulink Simulation Tools N. Wright
– Generic Simulink GUI– HEGO Simulink GUI
• Data Processor Tool R. Drane
• Summary R. Drane
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
50
Summary
Collected Data
Data Processor ToolProcess data collected for
calibration generation
MBC ModelModel calibration Surfaces
CAGEGenerate initial calibrations
SpreadsheetsGenerate initial
calibrations
Simulink Simulation ToolsRun simulation models with all
collected data sets
Tweak Calibration
Data Processor ToolConduct validation analysis on
output from simulink models
Collect Sign-Off Evidence Data
Not OK
OK
26
Nick Wright and Robin Drane The MathWorks International Automotive Conference, June 2004
51
Question Time
Any Questions
???