Tools for testing control systems
Report 2005.TL.6981, Transport Engineering and Logistics.
This report gives an overview of the various methods used for testing
control systems: hardware methods, CPU response methods, logic analysis
and simulation methods. This last method is the most suitable for testing
modern multi variable, non-linear control systems; therefore the focus
will be on this specific testing method.
Four types of simulation methods are distinguished: simulation, emulation,
real time control and prototyping. Emulation, the use of a real control
system with a simulated physical system, is the principal subject of this
work. This way of testing offers greater accuracy than simulation and it
is usually cheaper than prototyping. Other advantages include the
possibility to test the control system earlier in the development process,
the possibility to test the control systems response in situations that
are difficult, dangerous or impossible to reproduce with the real system
and the ability to use emulation for training purposes.
In order to create an emulation model, a design methodology has to be
applied. Different life-cycle models are available for the development of
software. The waterfall model, spiral model and object oriented life cycle
model are discussed; the most suitable model for building emulation models
is identified. A number of requirements have to be met to build a good
emulation model. Whether a model is 'good' is determined by testing:
model verification and validation. Attention will be paid to these
subjects, together with calibration: the fine-tuning of a model.
Finally, five applications of emulation in different domains are
discussed; emulation used in the field of container terminal design,
airport baggage handling systems and the design of Automated Guided
Vehicles (AGV) systems.
Reports on Transport Engineering and Logistics (in Dutch)
, TU Delft