scientific article; zbMATH DE number 17015

From MaRDI portal
Publication:3973700





[For the entire collection see Zbl 0705.00029.] The author refers to the problem of optimizing system maintenance. The proposed model is a semi-Markov process in which the control variables are \(G_{ij}(t)\) (the pmf for the time to next planned maintenance) and \(\bar\alpha_ i=(\alpha_{i1},\dots,\alpha_{im})\) the strategy (the probabilities to reach the proposed target state from state \(i\) after maintenance). The main quality index is: \[ V^*=\lim_{t\to\infty} V(t)/t), \] where \(t\) is time and \(V(t)\) a mean measure of the process on interval \([0,t)\). The article's main part consists of a theorem stating that \(V^*\) is obtained for degenerate, that is deterministic, \(G_{ij}(t)\) and \(\bar\alpha_ i\). There is no demonstration of the theorem but the reader is often referred to the references (previous papers of the same author, unfortunately given without an address). The second part of the article is dedicated to an example of application. At maintenance the target state is supposed to be (as stated in the theorem) only the first one, \(\bar\alpha_ i=(1,0,0,0)\) and the intervals to next maintenance activities are also supposed to be deterministic: \(u_{11}\), \(u_{21}\), \(u_{31}\), \(u_{41}\). After some calculations, the author derives a formula that can be differentiated with respect to \(u_{i1}\) \((i=1,4)\) to determine the optimum of \(V^*\). The frequent appeal to the references makes the paper difficult to follow.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3973700)