Importance and sensitivity analysis in dynamic reliability (Q631475)

Consider a piecewise deterministic Markov process, characterized by different input data depending on some parameter \(p \in P\). The problem is to compute the first-order derivative with respect to the parameter \(p\) of some functional of the process, cumulated on a finite time interval \([0,t]\) and its asymptotic value per unit time. Examples are cumulated (production) availability, or the mean number of failures on some finite time interval. The computation of the first-order derivative of the functional with respect to \(p\) is obtained through a probabilistic counterpart of the adjoint state method from the numerical analysis field. Some examples are given.