چكيده به لاتين
Hybrid systems consist of both discrete and continuous parts. These two parts work together continuously. The computational part receives environmental information, and then it sends the required commands to the physical section after performing the necessary calculations to modify their operations. This cycle will continue in hybrid systems until they reach their goal. Today, hybrid systems are widely used in a variety of areas, including health monitoring, automated transportation, and critical infrastructure of organizations. Occurrence of the failure in these systems can be harmful, due to the sensitivity of the application of these systems.
In each of the various components of the hybrid system, there can be one or more faults, and at any time, these faults can be activated and their effects will be propagated to other components, and it will lead to catastrophic events. Due to the sensitivity of the application of these systems and the complexity of their structures, it is necessary to design a fault propagation model to observe the propagation of the effects of a fault in a component to other components, before the construction of the system. It is also possible to identify critical components of the system. Therefore, designing a model for fault propagation in hybrid systems is essential. The fault propagation model should have the following features: design simplicity, scalability, comprehensiveness, and time-saving.
In this study, a fault propagation model for hybrid systems is presented. Fault propagation between the various components of a system is a result of data and non-data dependencies and the execution order of the components. In the proposed model, data dependency graph, non-data dependency graph and the execution order graph are extracted based on the system topology and behavior. Then the modeler considers the probabilistic terms related to the behavior of components. Then a model of stochastic activity networks is extracted using the graph conversion rules presented in this study. Finally, the system is monitored and evaluated by injecting various faults in the system model. Compared to previous works in this area, the proposed model has properties such as high-comprehensiveness for applying to any system, high scalability, and design simplicity. The final simulation results show that with this model we can evaluate various parameters such as the failure behavior of a component in the presence and absence of another component’s fault, the probability of the failure of the component at a specific time, the reliability of the entire system, and so on.
