Buch, Englisch, 392 Seiten, Paperback, Format (B × H): 170 mm x 240 mm, Gewicht: 787 g
Reihe: Luftfahrt
Buch, Englisch, 392 Seiten, Paperback, Format (B × H): 170 mm x 240 mm, Gewicht: 787 g
Reihe: Luftfahrt
ISBN: 978-3-8439-5627-7
Verlag: Dr. Hut
Integrated Modular Avionics (IMA) are the state-of-the-art approach for integrating mixed-criticality applications on shared hardware in aircraft. While IMA allow avionic systems to be compact, lightweight, power efficient, and cost-efficient, the number of integrated functions, future upgrades and customization of avionic systems are introducing complexity and remain cost-drivers.
Plug&Fly Avionics are a novel IMA approach that minimize manual configuration and reduce development effort by introducing self-adaptation on platform level. To do so, Plug&Fly Avionics implement an online model-based process that mimics the development process of ARP4754A and safety assessment methods of ARP4761. This work provides a run-time assurance architecture that addresses how self-adaptive Plug&Fly avionics can autonomously establish and ensure safe execution of hosted applications.
A methodology is presented in this thesis to allow the autonomous adaptation of applications into redundant, fault-tolerant realizations. Based on pre-supplied model-based application specifications that include safety requirements, a redundant design is determined using constraints programming.
To safeguard configurations put into live operation, a more detailed model-based safety assessment is performed prior to accepting the configuration. This ensures that only configurations that comply with the safety requirements of the respective application are executed in the platform. The safety assessment method presented in this thesis is based on failure propagation models, which describe the failure behavior of application parts. These models are compiled into Binary Decision Diagrams (BDD) for the exact determination of failure probabilities. A human-readable safety artifact in form of structured fault trees is synthesized from these BDDs to provide explanation and enable monitoring by human operators.
