Toward the optimal selective maintenance for multi-component systems using observed failure: applied to the FMS study case.

Several industrial systems are required to carry out a sequence of missions in which the maintenance activity is allowed only between two consecutive missions. In such case, the selective maintenance strategy is widely adopted, which aims to select the best components to be maintained, taking into account the constrained maintenance resources. The suggested maintenance selective resolution methods in the literature are based essentially on theoretical analyses and estimated parameters, which does not necessarily reflect the reality. In this study, we propose to combine techniques based on data mining approach to the selective maintenance of a multi-component system, using the maintenance data collection. For this purpose, the similar components are clustered, as a first step, and the age degradation coefficient of similar components is computed by using the K-means clustering algorithm. Next, we propose a mixed integer programming model that uses as input data the components degradation coefficients of the first step, in order to decide which components need to be replaced first. After, using the Apriori algorithm, the frequent sequences of maintained component are extracted, which provides us additional knowledge to predict the next maintenance activity. Finally, the proposed approach is assessed through a real-world case study of maintenance data of a flexible manufacturing system. [ABSTRACT FROM AUTHOR]