Safe reliability assessment through probabilistic constraint reasoning

Reliability quantifies the ability of a system to perform its required function under variable conditions. The adequate functioning of a system is often represented by inequality constraints, and its reliability is the probability that such constraints are satisfied, given the uncertainty the variables are subject to. Since this computation is very hard, namely when the systems are modeled with nonlinear constraints, traditional methods adopt a number of approximations, thus computing a value that may be far from the exact one. Moreover, these methods do not provide any guarantees regarding the correctness of the computed results. In this paper, we use the probabilistic continuous constraints framework to efficiently compute safe bounds for the reliability of a system, and illustrate it on a number of representative examples. © 2015 Taylor & Francis Group.