Establishing a Data Farm to Harvest Quality Information

摘要

To obtain useful reliability information from maintenance and inspection data once it has been collected requires effort. In the past, this has been referred to as data "mining", as if the data can be extracted in its desired form if only it can be found. In contrast, this paper proposes data "farming", and describes the seeds that are necessary to harvest the best possible crop of reliability information. The CCPS equipment reliability database project provides valuable lessons on how to "farm" rather than merely "mine" data. The CCPS work processes for establishing failure modes, populations to track, event data to collect, and implementation are all reviewed. Attention is given to knowing up front the data objectives and the quality of information desired. Also, the treatment of equipment surveillance periods turns out to be a critical variable for data quantity and quality; reasons for this and approaches to take are discussed. It will be seen that the quality and continuity of information that can be derived is much greater when the data sources can be "farmed" rather than "mined".