Abstract

The successful implementation of fault diagnosis and failure prognosis algorithms to safety critical systems requires the definitions and applications of mathematically rigorous modules. These modules, including data preprocessing, feature extraction, diagnostic and prognostic algorithms, performance metrics definition, and a fault progression model, form an integrated architecture for system health monitoring and management. In these modules, the fault progression model is critical to detection of incipient failures as early as possible with predefined specifications and prediction of the system's remaining useful life accurately and precisely. This paper considers an oil cooler bearing of a helicopter and proposes a methodology for fault detection and failure prognosis, in which data pre-processing, feature extraction and fault progression modeling are discussed in detail. Experimental results are presented to verify the proposed methodology and fault progression model. © 2010 AACC.