Clustering condition-based maintenance activities for multi-state systems with imperfect prediction signals

Industrial organizations are constantly looking for new strategies to improve their performance. Optimizing maintenance and choosing maintenance strategies play a critical role in the performance of any industrial system. Nowadays, most sophisticated systems are equipped with condition monitoring tools such as sensors and computing devices, which draw a lot of attention towards the condition-based maintenance. However, condition-based maintenance policies have not been implemented in practice. Joint execution of maintenance activities on multiple components is known as clustering of maintenance activities, which significantly reduces the costs of multi-unit systems. This paper developed a clustering model of maintenance activities for multi-state systems to reduce costs. Each degradation level can be represented by a state, making the system a multi-state system. The system releases signal each time its condition deteriorates and can be estimated with some uncertainty. The influences of parameters are analysed and it is determined that which type of clustering causes the most cost savings.

Keywords:

condition-based maintenance , clustering , imperfect prediction signal , multi-state system , delay time