Extraction and exploitation of R&M knowledge from a fleet perspective

The availability of failure data, condition monitoring data and operational data for large sets of homogeneous products (in the following, referred to as a fleet) motivates the development of new analytical methods to extract and exploit reliability and maintainability knowledge on a fleet level. In other words, by clustering products into fleets based on categories of interest (e.g. subsets of assets belonging to the same customer or installed in the same region or operating for the same industrial application) the fleet-specific usage and degradation profiles can be extracted by statistically analyzing the operational, monitoring and failure data of the assets belonging to the selected fleets. In this paper we describe the process of extracting reliability knowledge (in terms of usage and degradation profiles) at the single product level and at the fleet level. The extracted knowledge is then exploited to determine the remaining useful life of a unit in the fleet. In this context, a simple but an effective data-driven prognostic algorithm is proposed. The algorithm makes use of the knowledge extracted at a fleet level to enhance its forecasting capabilities.