System Probability Distribution Modeling Using MDDs
Mitchell Aaron Thornton, Theodore W. Manikas, Stephen A. Szygenda, Shinobu Nagayama · 2014
Multi-Valued (MV) fault trees can be used to represent a variety of probability distributions characterizing system-related events. Representing MV fault trees in the form of multiple-valued decision diagrams (MDD) provides a means for representing overall system probability distributions and are constructed from structure functions. MDD edges are annotated with component probability values and allow for computation of overall system probability values. MDD 'phantom' vertices are presented to overcome inaccuracies introduced by the 'rare event approximation'. Additionally, a method that allows continuous probability distributions as MDD edge annotations is described. Experimental results are provided that illustrate the viability of the method.