Hazard Analysis for Reconfigurable Real-time Systems

Self-optimizing mechatronic system often are applied in safety-critical environments. They have to satisfy high safety constraints. This requires a rigorous analysis. Even though the system does not contain any design errors, hazardous situations may be caused by random errors during run-time that happen, e.g., due to the wear of physical components. For analyzing the effect of these errors on the system and the probability of their occurrence we apply hazard analysis.

In the course of this PhD project, we develop a hazard analysis that addresses the properties of self-optimizing mechatronic systems. Therefore, the propagation times of failures and the effect of reconfiguration on the failure propagation are taken into account. The failure propagation models which are used for the hazard analysis are generated automatically from the behavior models of the software. This eliminates human mistakes from the creation of these models.

The hazard analysis is extended to a risk analysis that is executed during run-time. Risk is the combination of the probability of a hazard and the severity of the the hazard's consequences. The run-time risk analysis ensures that reconfigurations do not result in system states that violate the risk level of the system. This also applies to the case when system constraints like the severity change during run-time.

Contact: Claudia Priesterjahn