Towards Risk-Based Autonomous Decision-making with Accident Dynamic Simulation

Supervised dynamic probabilistic risk assessment: Review and comparison of methods

“…Supervised DPRA, however, attempts in different ways to reduce the computation complexity and solve the state explosion problem. Computational complexity analyses were performed on the supervised DPRA methods, where their worstcase time complexities were defined and compared to the baseline exponential complexity defined in [50]. More specifically, the algorithms for each of the 52 supervised DPRA paper were analyzed in terms of the number of operations performed, yielding computational time complexities.…”

“…Asymptotic computational complexity analyses as described above were performed here for the reviewed supervised DPRA methods, considering the worst-case approach, as shown in Table A.2. The objective is to compare the supervised DPRA methods to each other and to the baseline computational complexity of DPRA, which is exponential [50]. Furthermore, we compare the methods in traceability (i.e., if the undesirable consequences are reachable from the initial conditions) and if they cover the full state-space of possibilities (i.e., a complete search of the state-space).…”