| Summary: | Nowadays, the developed diagnostics models and software are not capable of locating the root cause of an emerging malfunction, or in other words the responsible component, while the vehicle is up and running. In most cases they are solely able to provide the driver with indications that a fault has been detected within a group of components. Subsequently, it is unavoidable that the vehicle returns to the workshop for a number of standardized tests to be performed, in order to evaluate the condition of the potentially faulty components. The new era in combustion engines and the attempt to fully incorporate closed-loop combustion control can facilitate the diagnostics procedure and especially the process of fault isolation. By harnessing signals from both real and virtual sensors, it can be feasible to diagnose or even prognose faults, averting the return of the vehicle to the workshop. Moreover, the down-time of the vehicle, can be radically decreased, since there will be an indication on which components to focus. Taking into account the fastpace steps and improvements on the respective hardware, such as sensors, one can understand that this endeavour can actually be successful in the future. In the spectrum of this thesis it is assessed whether or not fault detection and isolation can be achieved, through comparison of sensors’ output signals for a number of engine parameters to a stored set of nominal values for these parameters (reference values). Towards that goal, virtual sensors have been developed with the aid of measurement data, in order to increase the reliability of the system. Subsequently, a network of dependencies between parameter values and consequent malfunctions has been constructed, in the form of flowcharts, rudimental for fault isolation. In addition to that and despite the fact that no finalized production code for the model is provided, pseudocode charts have been created as well. Finally, significant effort was made to derive precise tolerances for the reference values, as this is of great importance for the results of the diagnostics model.
|
|---|
