B.4. Details of convergence of metrics over generations 209 Inferred COVID-19 infection risk fault tree Human error HE Existence of vulnerable workers VW AND COVID-19 infected worker on site TE COVID infected surface IS Mutual use of vehicles MV COVID-19 infected object used by the team Existence of COVID-19 IT COVID-19 unknown modes of transmission UU OR AND OR IW COVID-19 infected worker joining the team Transmission modes COVID-19 Airbome CA Physical proximity PP Transmissibility of COVID-19 pathogen IE1 IE2 IE3 Figure B.3: Inferred COVID-19 infection risk fault tree after applying FT-MOEA, source: Bakeli, Hafidi, et al., 2020. the extreme values in that generation for a given metric, and a white dashed line indicating the mean value of the metric. Figures B.4(a) and B.4(b) illustrate the convergence of ωd. By using the multiobjective function (m.o.f.) sdc, we observe higher variance compared to the m.o.f. d throughout the generations. This is due to the fact that some FTs are Pareto optimal in other aspects, e.g., FTs with a small size, which often have higher error. In contrast, Figure B.4(b) shows less variance, indicating that FTs within a generation have a similar error based on the failure dataset (ωd).
RkJQdWJsaXNoZXIy MjY0ODMw