Dependability for systems with a partitioned state space: Markov and semi-Markov theory and computational implementation (Q1335296)

The monograph is a compilation of recent results of the author on several aspects of sojourn times in Markov or semi-Markov processes with a finite, partitioned state space \(S\). The chief motivation is that the partition classes \(A_ i\) may be associated with different levels of functionality of a system, and throughout the emphasis is on deriving explicit formulae pertinent to reliability studies. Each of the main theoretical developments is applied to one or more power engineering problems taken from the literature, and computational issues are discussed in some detail. Several program codes are given, notably for the MATLAB software tool. Concerning the mathematics: (Semi-)Markov processes on \(S\) are considered which are either irreducible or absorbing. Let \(I_{A_ i,m}\) be the \(m\)th sojourn interval for the process in class \(A_ i\) (\(=\emptyset\) if there are fewer than \(m\) sojourns). The main variables of interest are \[ T_{A_ i,m}(t)= \text{meas}(I_{A_ i,m} \cap[0, t]),\;0<t\leq\infty, \quad\text{ and } \quad M_{A_ i}(t)=\sup\{ m>0:\;T_{A_ i,m} (t)>0\}. \] The latter is the number of visits to class \(A_ i\) up to time \(t\), and may be allied to the tally of certain repair events. Results are obtained for various joint distributions of these variables, in both discrete and continuous time. The proofs involve substantial manipulation of partitioned matrices, transform techniques, renewal arguments, etc. Despite the unwieldiness of some of the expressions, the account is orderly and clear.