A new problem for selective maintenance considering bi-objectives, repairperson assignment and k-out-of-n systems

Abstract

This dissertation deals with the maintenance optimization problem in a multicomponent system, which should undergo maintenance actions between two consecutivemissions, preparing itself for the next mission. Due to time, budget and resource limitations,top-level actions cannot be performed on all components and therefore, a subset ofcomponents and actions should be selected for the objective optimization. Most of theexisting models to tackle this kind of problem do not involves complex systems or, when theydo it, they consider only one objective to be optimized. To study the establishment ofproblems that consider complex systems, multi-objective approaches and repairpersonassignments, this work proposes a new non-linear binary model that models the bi-ObjectiveSelective Maintenance and Repairperson Assignment Problem on k-out-of-nsystems (biOSMRAP:k-out-of-n). Its modeling is discussed, and three algorithms are proposed for theproblem solving: a full enumeration algorithm, a metaheuristic and a matheuristic, these lasttwo based on the Adaptive Variable Neighborhood Search. Two instances were tested, oneartificial instance and the other from the literature, and a sensitive analysis was conducted tounderstand the problem behavior. Both approximated algorithms were solid, supported bygood values for the metrics used.