Abstract
To realize a predictive maintenance strategy, it is necessary to maintain high accuracy in failure prediction. For this, it is essential to update a predictive maintenance system continuously, and previous study proposed an evolution-oriented predictive maintenance framework. In the proposed framework, the knowledge on maintenance and the data of operation and maintenance are accumulated in a dedicated repository and reused in the predictive maintenance system to continuously improve failure prediction accuracy and support maintenance engineers' activities. However, in this framework, the essential knowledge and data to be accumulated remain unclear. To address this issue, this paper aims to design repository, mainly targeting the support of maintenance engineer' activities. We clarified the requirements for the repository by collecting and organizing knowledge in maintenance activities based on interviews with engineers and analysis of current maintenance activity record. Based on the requirements, this study designed repository, that is, clarified knowledge representation, data representation, and the representation of the relationship between the knowledge and data. To verify the developed model, this study implemented a part of maintenance knowledge base based on the defined repository and demonstrated the maintenance engineers’ activity support by utilizing the repository. As a result, the knowledge extraction has succeeded, and this study clarified that the repository is functionable for engineers’ support.
