This paper presents a new technique for dynamic scheduling of industrial maintenance based on the use of genetic algorithms. This will allow to propose the best time to schedule each maintenance activity in terms of technical and economical considerations and also taking into account the production planning and operation conditions of the equipment or industrial plant. The scheduling module is in charge of collecting and scheduling these maintenance activities according to their execution-priority and also, taking into account the production planning of this equipment as well as minimizing the overall maintenance cost and considering the best operation conditions for carrying out the maintenance activities. Besides, at any time, either the list of maintenance activities or their priorities or the economical or operational considerations or the resources available for making the maintenance may change, forcing a re-scheduling of the maintenance planning. Thus, the scheduling is modeled by a dynamical multi-criteria optimization dealing with non-linear relations and restrictions and considering time-varying availabilities and resources. Moreover, maintenance activities durations and costs are usually imprecisely known or may vary along certain time intervals. Thus, fuzzy numbers are used to quantify these durations and costs because they provide more available information than just using a simple number. In addition, a more flexible scheduling is obtained and it can be assured the accomplishment of every requirement.
Keywords: Dynamic scheduling, genetic algorithms, heuristics, multi-criteria optimization, production planning
16th International Euromaintenance 2002. Helsinki (Finlandia). 3-5 June 2002
Published: June 2002.