< Back to previous page


i-RCAM: Intelligent expert system for root cause analysis in maintenance decision making

Book Contribution - Book Chapter Conference Contribution

The increasing adoption of maintenance information management systems for maintenance decision support by industry has facilitated the collection of large volumes of maintenance data. Apart from enhancing maintenance decision support in aspects such as task planning or resource allocation, the data could assist decision makers identify the focal root causes of recurrent equipment failures. In this way, more effective strategies may be formulated and targeted at these focal causes. Despite the increased adoption of maintenance information systems and as such, availability of maintenance data, few techniques so far developed leverage on the maintenance data for decision support in root cause analysis. A particular focus in this regard relates to application of techniques for data mining such as association rule mining. In particular, association rule mining is attractive in the sense of analyzing failure associations embedded in the maintenance data. Thus, this study proposes a methodology for enhancing decision support for root cause analysis in maintenance decision making. The methodology leverages on two association rule mining algorithms - Apriori and Predictive Apriori. Moreover, the methodology incorporates a data standardization step, whereof standard terms and vocabulary are adopted from the ISO 14224 and used for standardizing the equipment failure descriptions. Thereafter, the standardized descriptions are applied as input to an association rule mining framework from which important failure associations are extracted and validated by experts for relevancy. After which, the extracted failure associations are used to generate causal maps, and from the maps, the focal root causes of the equipment failure are identified. The added value of the proposed methodology is demonstrated in the application case of thermal power plant maintenance data.
Book: 2016 IEEE International Conference on Prognostics and Health Management
Number of pages: 7
Publication year:2016
Authors from:Higher Education