Maintenance policies for systems with condition monitoring and obvious failures

Structured maintenance policies on interior sample paths

“…As an initial step in this direction, future work could consider the same framework considered here, but with obvious failures and investigate how the necessary conditions can be strengthened to establish the same structured policy results along ordered sample paths. For this problem, in which observing no failure in a particular period yields imperfect information regarding the underlying deterioration state, structural results have been examined along straight lines of information states (Ohnishi [4], Maillart [3]), but not sample paths.…”

Robust Control of Partially Observable Failing Systems

“…In the maintenance optimization literature, one of the key structural properties to establish is convexity of the value function (see, e.g., Dayanik et al 2008, Kim and Makis 2012, Maillart 2006, Makis 2008). In the current setting however, it is not clear that the robust value function W defined in (14) will preserve such a convex structure in the presence of positional and transitional model ambiguity.…”

“…In the condition-based maintenance literature, there has been a long and rich history of establishing structural properties of optimal maintenance policies (see, e.g., Ross 1971, Makis and Jardine 1992, White 1978, Elwany et al 2011, Maillart 2006, Kurt and Kharoufeh 2010, Ulukus et al 2012. Although a lot of research has been conducted in this area, to our knowledge, this paper constitutes the first paper reporting structural results for maintenance optimization in the presence of model miss-specification.…”

“…A common modeling approach is to represent system degradation as an unobservable Markov process that can only be indirectly or partially observed using stochastically related signals collected through online condition monitoring (see, e.g., Dayanik and Gurler 2002, Dayanik et al 2008, Kim and Makis 2013, Maillart 2006. From the perspective of real-world implementation, this modeling approach implicitly assumes that the decision maker is able to accurately and completely specify the full joint distribution governing the real-world degradation/failure dynamics as well as the condition-monitoring observations.…”

Optimal maintenance of a series production system with two multi-component subsystems and an intermediate buffer

“…A crucial part of the MDP model is the relative cost function that formulates the relative cost of a single step in the long-run decision process [14]. For the investigated maintenance optimisation problem, the relative cost function is given by:…”