Data-driven approach to equipment taxonomy classification

Abstract

A standardized taxonomy enables asset-intensive industrial organizations to systematically measure and track efficiency and performance of assets at different levels in an asset hierarchy. Having a well-structured taxonomy also allows companies to take advantage of emerging data-driven technologies, such as prognostics and health management (PHM), through enabling straightforward mapping of assets to analytical content specific to equipment commonalities, e.g., failure modes. However, the complexity and use of equipment taxonomy and coding structures in maintenance management systems vary widely for different organizations. This paper describes a data-driven approach for identifying equipment taxonomy from equipment records in maintenance management systems. The approach combines machine learning-based and rule-based methods into a hybrid man-in-the-loop workflow, which enables rapid and consistent mapping of equipment into a standard taxonomy. A case study is presented to demonstrate the performance and challenges of the proposed approach on equipment taxonomy classification.