As indicated by the state-of-the-art, models have to be identified that characterise behaviours at
an appropriate level of abstraction, particularly capturing the dynamic aspects in a discretized,
quasi-static or equilibrium scheme. The main source for technical risks in the project lies
in the capability of all partners to build appropriate models describing the essential features
of selected automotive subsystems for diagnosis.
One of the risks stems from the fact that the subsystems considered may
involve components with complex temporal behaviour and feedback in a way
that limits applicability of available techniques. The researchers involved
are investigating simpler modelling approaches by reconstructing the known
brain-storming procedure in FMEA meetings.
Another risk factor related to dynamics is the real-time behaviour of
diagnostics required for on-board diagnosis. Besides techniques for model
compilation, this has to be addressed by appropriate methods for model
abstraction and focused reasoning, and may involve as trade-off
between speed and completeness of the diagnosis. The underlying problem of
real time monitoring and reaction in rapidly changing systems can be a
safety problem and is tackled in present automotive subsystems by offering a
"limp home" option where possible.
Furthermore, the involvement of (control) software in the subsystem
may impose restrictions to the competence of the diagnostic systems, if
modelling of the functionality of the software is required and beyond the
state-of-the-art.
It is not obvious at this time, whether and how the use of wear models can
be integrated with diagnosis based on functional and/or behaviour models.
As far as wear models are based on statistical information, this may be
integrated on the basis of utility theoretic methods for diagnosis.
Beyond this, the project is expected to produce ideas and empirical material
that contribute to an understanding of the relation of wear models and
behaviour, rather than starting from a particular approach.
These risk factors could have effects on the planned targets in such
a way that the selected car types and their subsystems will not be
diagnosed completely. Thus only complex aggregates of these systems consisting
of electrical, electronical and hydraulic components will be handled and
demonstrated. The project may have to decide upon a proper selection of
subsystems and diagnostic tasks that can be expected to be successfully dealt
within the limited time frame of the project. Delivering a working solution
of a reduced set of subsystems has priority over a complete coverage
of systems that exists only on paper. It will be necessary, however,
to document the reasons for the selection of these subsystems and, even
more important. to identify the reasons why other systems appear not feasible
with the given methods and technical basis. This will be crucial for an
assessment of future research required and of the scope of applicability of
the delivered solutions.
