AN INTEGRATED CLUSTERING-DESIGN OF EXPERIMENT ALGORITHM FOR FAULT DIAGNOSTICS IN CONDITION BASED MAINTENANCE

A complete CBM system is composed of a number of functional capabilities: sensing and data acquisition, data manipulation, condition monitoring,health assessment/diagnostics, prognostics, and decision reasoning. In this paper, an integrated clusteringanalysis of variance (Clustering-AND VA) algorithm isproposed to improve diagnostics in CBM. This algorithm uses condition monitoring (CM) data jointly with system health data to construct some referenceconditions. The Clustering-ANOVA algorithm compares a new condition with all reference conditionsthrough ANOV A and calculates a degree of statistical similarity between new condition and reference conditions. The maximum of similarities between a new condition and reference conditions specifies the most similar reference condition and determine the status of the system. To show the applicability and usefulness ofthe proposed algorithm, a real case of rotating machinery vibration data in a petrochemical plant is used. Furthermore, sensitivity analysis is performed to show the validity and robustness of the ClusteringANOVA results.

Keywords:

Condition-based maintenance , diagnostics , clustering , ANOV A