Probabilistic Safety Risk Analysis in Complex Domains: Application to Unmanned Aircraft Systems
James T. Luxhøj, Michael Morton · 11th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference · 2011
Aviation is a complex domain characterized by low probability, high consequence events with scarce data; hence, safety risk modeling is particularly challenging. There continues to be a persistent need to develop the analytics to capture both the explicit and implicit risks inherent in such domains. A Value-based Time-phased Bayesian Network (VTBN) extends a conventional Bayesian Network (BN) by including features from a Dynamic Bayesian Network such as temporal risk factors that are enhanced by Multi-Attribute Value Theory (MAVT). The proposed VTBN integrates the quantitative analytic constructs of BNs and MAVT with the qualitative formalism of a structured hazard taxonomy. The enhanced methodology provides a framework for the systematic inclusion of the explicit risk inherent from the BN and the non-apparent implicit risk in the BN that exists in large complex systems. Preliminary modeling results suggest that VTBNs offer promise for advanced risk assessment, particularly for Unmanned Aircraft Systems (UAS) where civilian data are especially sparse. In this paper, the analytic constructs of a VTBN are demonstrated with an application of safety risk modeling to aid in the prioritization of a portfolio of mitigations for a futuristic UAS scenario.