Distributed Multi-Algorithm Diagnostics and Prognostics for US Navy Ships

George D. Hadden, Peter Bergstrom, George Vachtsevanos, Bonnie Holte Bennett, and Joe Van Dyke

Honeywell and its teammates (PredictDLI, Knowledge Partners of Minnesota, the Georgia Institute of Technology, York International, and WM Engineering) have developed a distributed shipboard system to perform diagnostics and prognostics on mechanical equipment (e.g. engines, generators, and chilled water systems) for the Office of Naval Research (ONR). This Condition Based Maintenance (CBM) system (called MPROS for Machinery Prognostics/Diagnostics System) consists of MEMS and conventional sensors on the machinery, local intelligent devices (called Data Concentrators), and a centrally located subsystem (called the PDME for Prognostics, Diagnostics, Monitoring Engine) which is designed so that it can run under shipboard monitoring systems such as ICAS (Integrated Condition Assessment System). The system uses an open, object-oriented approach with a welldefined API so that additional diagnostic and prognostic algorithms can be incorporated in a "plug and play" manner. MPROS includes and augments periodic vibration analysis by collecting data continuously from vibration and other sensors, including temperature, pressure, current, voltage, and others. These data streams are integrated as necessary in the Data Concentrators (data fusion). Individual prognostic and diagnostic algorithms can reside in either the Data Concentrators or the PDME. A second level of integration (Knowledge Fusion) occurs in the PDME. At this level, using both Dempster Shafer Evidence Combination and a mechanism to fuse time-to-failure estimates, the conclusions of the diagnostic and prognostic reasoning algorithms are fused to yield the best possible analysis.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.