Вы находитесь на странице: 1из 4

INDUSTRY DEVELOPMENT PROCESSES

ECU FUNCTION DEVELOPMENT


OFFLINE CALIBRATION AND VALIDATION
As a manufacturer of high-tech engines, Audi relies on the in-house development of central engine ECU functions.
Depending on the application and project specifications in individual cases, the respective ­functions are developed
either in Ascet or in Matlab / Simulink. Prototyping in the vehicle or on the test bench is handled by systems supplied
by dSpace and Etas. For calibrating production engine ECUs, Audi deploys the Inca measuring and calibration sys-
tem. The authors from Audi, Etas and Matis show how this tool enabled the calibration and validation of function
models for the engine management on the PC in the Simulink environment.

As regards the calibration and validation simplifies the precalibra­tion of ECU func- offline simulations on the PC with Inca,
of new controls, Audi’s objective is to tions but also facilitates – in one and the Etas was asked to integrate Inca directly
deploy Inca as early as during function same Inca environment – a direct compar- into Audi’s Matlab/Simulink environ-
development with Matlab/Simulink. This ison of the behavior of functions in the ment, 1. The resulting integration pro-
approach has several inherent benefits. ECU, the prototype, and the function vides function developers with the numer-
On the one hand, users are able to work model. ous special tools incorporated in Inca –
in their familiar calibration environment. such as virtual measuring instruments,
On the other hand, the exchange of mea- calibration editors, analytical tools for
INTERFACING INCA AND
surement and calibration data between measurement and calibration data, plus
MATLAB / SIMULINK
individual development steps and the its various triggering options – all of
final production calibration is seamless With the intent to enable the calibration which are absent from the generic
and entirely transition-free. This not only and validation of functions directly during ­Matlab/Simulink environment.

36
AUTHORS

DIPL.-ING. (FH) JOHANN GABLER DIPL.-ING. DANIEL KÖHLER DIPL.-ING. THOMAS HEINZLER DR. ULRICH LAUFF
works in the Development Area for is Project Manager and Function is Development Engineer at ­M atis is Technical Editor-in-Chief for
­Rapid Prototyping Tools in the ­D eveloper in the Software and Deutschland GmbH in ­M unich the Measuring, Calibration and
­S oftware and Function Development ­F unction Development Section for (Germany). He is a dedicated ECU Diagnostics Application Areas
Section for Powertrain at Audi AG Powertrain at Audi AG in Ingolstadt ­D evelopment Partner of Audi AG. in the Marketing Department of
in Ingolstadt (Germany). (Germany). Etas GmbH in Stuttgart (Germany).

MODELING GUIDELINES nique fosters a common understanding and signal values can be automatically
among developers, it also promotes the extracted from the Simulink model and
In order to implement the most seamless exchange of complex functions among transferred to Inca.
connection possible between calibration project participants. At Audi, the desig-
tool and simulation environment, adher- nations of ECU parameters and signals
INCA-SIP
ence to Audi’s specific modeling guide- are centrally managed by means of an
lines was required. Audi uses standard- Automotive Data Dictionary (ADD). In The integration of Inca was implemented
ized MBFS modules (Model Based Func- conjunction with the MBFS standard, the in the form of a Simulink toolbox.
tion Specification, an ASAM standard) to ADD name register unambiguously Installing the Inca-SIP toolbox equips
assemble ECU function models inde- defines measurement signals, character- Matlab/Simu­link with additional menu
pendently of development tools. While istic values, program maps, and axes. functions that are then used to establish
this ASAM-conformant modeling tech- Thanks to these prerequisites, parameter the connection between Inca and the

0 5 I 2 0 10 Volume 5 37
INDUSTRY DEVELOPMENT PROCESSES

1 Integration of Inca
and Matlab/Simulink

Simulink model, 2. As part of the ini- Simu­link, Inca permits the starting and vehicle. The data exchange with Inca is
tialization of a simulation, Inca-SIP gen- stopping of measurements, and their handled by an XCP interface provided by
erates – based on the analysis of the recording in the form of MDF files. Inca-SIP. This interface makes it possible
Simulink model – a descrip­tion of the Inca-SIP supports two simulation to run the simulation, independently of
calibration data in the ASAM A2L for- modes: The Fast Emulation Mode runs the Inca, on a sufficiently powerful computer –
mat, a hex file, and an experiment con- simulation at maximum speed. In the a capability that also supports the simula-
figuration in Inca. In the event of a new Real-Time Emulation Mode, the in-vehicle tion of ex­­tremely complex models on cen-
initialization of a simulation subsequent timing behavior of a given function is all tral servers.
to a functional modification of the but identical with that of the simulation. Inca-SIP enables Inca to automatically
model, the Inca objects are automati- Using this mode, the model can be record measurement data generated in
cally updated. With the model running calibrat­­ed with the same dynamics as in- Simulink. This measurement data can be
used as a reference for both the subse-
quent development steps and in-vehicle
road testing. Conversely, the simulation
can be stimulated with captured in-vehi-
cle Inca measurement data and the be­­
havior of a specific function analyzed in
the Simulink model. This practice facili-
tates the diagnosis of functional failures of
prototype or production functions in the
Simulink model.

APPLICATION EXAMPLE:
CYLINDER CHARGE DETERMINATION

As a first project at Audi, Inca-SIP is de­­


ployed for the offline calibration of the
cylinder charge determination based on
intake-manifold pressure 3. As the accu-
rate calculation of the fresh air mass
2 Inca-SIP throughout
the function development within the combustion chamber forms
process the basis for precise fuel metering, it

38
3 Cylinder charge determination based on intake-manifold pressure

directly influences fuel consumption and For this reason, a self-optimizing neural through the COM interface. Etas has been
emissions. Compared with alternative network was chosen for the mathematical offering Inca-SIP as part of its product
systems, the advantages of systems description of the system behavior. The range since mid-2010.
based on intake-manifold pressure con- complex model encompasses several hun-
sist of the acquisition of the load value dred ECU parameters and signals. The use
proximate to the combustion chamber, of Simulink and Inca serves to validate
the lower pressure loss in the intake function behavior, accommodate function­

THANKS
tract, and a higher degree of sensor ­al adaptations, as well as precision cali-
robustness. bration of the model.
On the downside, the process based on
intake-manifold pressure places high
OUTLOOK Special thanks to Oliver Schade, who was
demands on both control functions and
calibration. When used for complex For the next application step, Audi plans deeply involved in writing this article. He

engines, e.g., equipped with variable to deploy Inca-SIP for the automated data heads an Etas development team for custom-

valve control, charge-motion valve, and logging during function tests. This auto- er-specific software ­solutions and tool exten-

intake-manifold switching, nonlinear mation may be controlled by either Mat- sions in Ann Arbor, Michigan (USA).

effects may occur. lab/Simulink via M-scripting or by Inca

0 5 I 2 0 10 Volume 5 39

Вам также может понравиться