Frontloading through Virtualization Stay Real, Start Virtual
Ms. Shruthi Ananthachar Mr. Nithin Nath 1
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Increase in Control System Complexity − Pressure to differentiate through innovative functions − Many innovations are either safety-relevant and/or involve connectivity to the outside world − Restrictive legal and standard requirements − Cost and time-to-market pressures, high penalties for poor quality
Electronic fuel injection Cruise control
1975 2
Gearbox control Traction control Anti lock brakes Electronic fuel injection Cruise control
1985
Airbags Electronic stability control Active body control Adaptive gearbox control Adaptive cruise control Emergency call Gearbox control Traction control Anti lock brakes Electronic fuel injection Cruise control
1995
Tele-diagnostics Internet browser Apps Car-2-car communication Road trains Software updates Adaptive headlights Active steering Curve warning Stop and Go Lane keeping assistance Automated parking Collision mitigation Hybrid powertrain Emergency call Airbags Electronic stability control Active body control Adaptive gearbox control Adaptive cruise control Gearbox control Traction control Anti lock brakes Electronic fuel injection Cruise control
2005
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
2015
Frontloading through Virtualization Increase in Calibration Complexity SWComponent 1
SWComponent 2
SWComponent 3
AUTOSAR Interface
AUTOSAR Interface ‌
AUTOSAR Interface
Engine Management
AUTOSAR-RTE Basic Software Module (BSM)
Transmission Management
Microcontroller Abstraction ECU Hardware
Vehicle Motion Management
New layered standard architecture
Brake Management
Effort Number of calibrations
2002
2004
2007
25000
16000
12000
9000
4000 1997
2010
40000+
Increased inter-domain connectivity
100%
0%
2020
Engine control complexity 3
Battery Management
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
New functions and variants
Frontloading through Virtualization …leads to significant cost reduction
Simulation
Benches
Vehicles
Effort
Today – 70%
2015 – 40%
2015 – 20%
Today – 25%
Today – 5%
Reduce Cost for Function Development and Calibration
4
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
2015 – 40%
Frontloading through Virtualization Edit-Integrate-Validate-Update Validation Feedback Loop is a key towards efficiency
Develop Concept, Implement Concept, Calibration, Fine tuning
Integrate Functions with base SW, Simulation Environment
Edit
Update
Change specification, Debug SW, Change parameter
6
Integrate
Validate
Validate the function in Virtual or real environment
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Motivation
7
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Compressing the V-cycle
8
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Motivation
9
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Function development Frontloading
Validating new vehicle functions today‌
Reserve a test vehicle
Flash software or configure hardware for prototyping
Make required adjustments
11
Validate against requirements
Passed Failed
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Function development Frontloading
Validating new vehicle functions in the future‌
Virtual Prototype
Make required adjustments
12
Compile
Validate against requirements
Passed Failed
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
F4
F7 New developed functions, e.g. in ASCET or Simulink®
+
Automotive Production OS
+
Production Function Componentss (FCs)
+ + +
Measure Data Files (Stimuli) Plant Model
ECU FCs
OS
F Plant Model
F
Calibration Infrastructure
Solution − Create a complete virtual prototype to allow detailed validation − Execute virtual prototype on a standard windows PC − Experiment and calibrate with existing tools e.g. INCA
− Better quality models streamlines collaboration 13
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
ETAS INCA
MDF Stimuli
Frontloading through Virtualization Co-simulation during modeling
14
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Software development Frontloading
Taking Benefit of AUTOSAR: same SW under test in all environment
Application Software RTE Basic SW OS Virtual PC MCAL
15
OS
OS Real-time PC MCAL
OS Rea-time-PC MCAL ECU MCAL
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Real-time-PC MCAL
Frontloading through Virtualization Function development Frontloading
Virtual ECU
16
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Testing Frontloading
17
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization ETAS approach
18
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Calibration Frontloading
19
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Conventional Calibration Approach
Calibrator Workstation
I don’t believe in virtualization… I don’t think it even works! I prefer to work on a “real” target
Vehicle Environment
20
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Conventional Calibration Approach Users of Conventional Approach expect…
• Very high accuracy – Best trade-off between model-error & no. of training data. • Global engine models for calibration possible.
• Best model is built automatically. • Easy and descriptive rating of model quality – Only basic DoE and statistic knowledge needed • Many optimization features for calibration, e.g: multi-criteria (pareto) or fast cycleoptimization • Easy & flexible data import & export: xls, txt, csv, dcm … • Open Matlab interface for scripts, extensions or test bench connection. Model export e.g. to C, Matlab, Simulink, Excel, … 21
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Model & Automation based Calibration Approach
Model Based Calibration with ASCMO ® Injection
Consumption
Ignition
Torque
AFR
HC
EGR P-rail
NOx
Inlet-cam Outlet-cam
Soot T-exhaust Roughness
Guided Calibration with INCA-FLOW ®
Integrated solution for the calibrator
Advanced Simulation for Calibration Measurement Optimization 22
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Calibration Frontloading – model based approach
23
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Calibration Frontloading – model based approach
250 230
Scalability & re-use of global data • Varied Torque • Payload • Fuel Economy target • ….
Power
200 180 160 130 100
Engine Speed 24
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Pictures courtesy Ashok Leyland
Frontloading through Virtualization Calibration Frontloading – model based approach Polynomials or Neuronal Nets
Statistical machine learning methods
Principle:
Principle:
Search in a complete function space:
Search in a given class of functions (polynomial, neuronal net, ...) Fit the model parameter by experts and validation measurements
Automatic determination of the most likely function
Disadvantages:
Advantages
High flexibility without assumptions or expertise Gives local confidence interval (model variance) Robust against outliers
Limited flexibility & danger of over-fitting High expertise and assumptions necessary
Training Data
Training Data & Model Prediction
Model Prediction
Modelvariance & Validity
Modeling a complex 1-D signal with classical DoE-Models („Advanced Polynomials“) 25
25
Modeling a complex 1-D signal with new statistical machine learning methods
Public| ETAS India| 2014-05-29 | © ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Calibration Frontloading Use-Case Smoke Limitation Optimization on a Diesel platform Visual Optimization Method – Partial Online ECU software Hex +
A2L+
ES592
Full-Load Curve
Torque
Driver Demand Smoke Limit Curve
Engine Speed 26
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Calibration Frontloading Use-Case ECU software Hex +
A2L+
ES592
INCA FLOWÂŽ Sequence
INCA On-line Environment
27
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Calibration Frontloading Use-Case
DoE-Plan
Model training using measured data from DoE-Plan Mathematical Model CO2 Ordnung 3, Minimum bei (170Nm,30kW)
165 160 155 150 350 145
300 100
Systeminputs
28
250
80 200
60 40
150 20
Systemoutputs
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Optimising outputs based on the model
Frontloading through Virtualization Calibration Frontloading Use-Case
Model based & Guided Calibration reduces calibration effort by roughly 70% for this use case!
29
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Summary – Advantages
30
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.
Frontloading through Virtualization Stay Real, Start Virtual
31
Public| ETAS India| 2014-05-29 | Š ETAS GmbH 2014. All rights reserved, also regarding any disposal, exploitation, reproduction, editing, distribution, as well as in the event of applications for industrial property rights.