Accession Number:

ADA564389

Title:

A Dynamic Neural Network Approach to CBM

Descriptive Note:

Final rept. Sep 2009-Dec 2010

Corporate Author:

MIS2000 GLOBAL DEFENSE ELECTRONICS INC SOUTHFIELD MI

Personal Author(s):

• Zannini, Margherita
• Feldkamp, Lee
• Marko, Ken
• James, John
• Maccani, Larry
• Wiebe, Robert

Report Date:

2011-03-15

Pagination or Media Count:

74.0

Abstract:

This project was the continuation of an initial project regarding the use of Neural Networks as they related to Condition Based Maintenance of military vehicles. The overall objective of the project was to investigate the use of prognostic algorithms and neural networks as they pertain to powertrain systems by creating relevant faults in the engine operating conditions related to fluid temperature, pressure, and flow, which produce performance loss and impact the vehicle health. Our investigation was carried out on the military version of the Caterpillar C7, 7.2L 6 cylinder engine mounted on an Eddy current dynamometer at the dynamometer facility of the Mobility Group at the Detroit Arsenal. The engine is a diesel in-line with a waste-gated turbocharger Faults were introduced by altering the normal response of some electromechanical components of the engine control system. Known malfunctions were generated in such a way that the faulty condition could be turned on and off without actually exchanging malfunctioning components.

Descriptors:

• *DIESEL ENGINES
• *MAINTENANCE
• ALGORITHMS
• DETECTORS
• FAULTS
• MALFUNCTIONS
• MILITARY VEHICLES
• NEURAL NETS
• TURBOCHARGERS

Subject Categories:

• Cybernetics
• Logistics, Military Facilities and Supplies
• Reciprocating and Rotating Engines

Distribution Statement:

APPROVED FOR PUBLIC RELEASE