Combating Engine Failures in Tanks - using artificial intelligence to find problems - Brief Article

USA Today (Society for the Advancement of Education), June, 1999

Artificial intelligence could boost the Army's battle against budget crunching as well as aid battlefield readiness by diagnosing engine problems in tanks before costly repairs are needed. Researchers at the Department of Energy's Pacific Northwest National Laboratory, Richland, Wash., are developing Turbine Engine Diagnostics Using Artificial Neural Networks (TEDANN). The technology uses diagnostic engineering, artificial neural networks, and model-based deck sion algorithms to predict failures and abnormal operations in the M1 Abrams main battle tank's turbine engine.

Currently, sensors aboard a tank indicate only if the engine's operations are in or out of tolerance--that is, if a problem does or doesn't exist. TEDANN monitors various engine conditions continually and tracks potential deviations from normal operations. Through TEDANN's predictive diagnostics, maintenance personnel could be alerted to adverse conditions as they develop, potentially reducing expensive engine failures.

The Army schedules periodic engine checks on M1 tanks that require the engine to be removed from the hull. Utilizing TEDANN, these would be performed automatically during normal operations. Mechanics would know if an engine needs maintenance before its scheduled service date or if it is in good health and doesn't need servicing.

TEDANN utilizes data from 32 existing sensors on the gas turbine engine and 16 new sensors that have been added using a wiring harness. These sensors continually monitor engine performance and relay the information back to a computer processor for analysis, which is done using artificial neural networks. These networks process information the way a human brain does--learning by example. For TEDANN, they are configured to model the behavior of normal engine performance and to recognize deviations.

A mechanic is able to access TEDANN's analysis using a laptop computer. He could monitor the sensor data that has been collected (for instance, engine temperatures, fuel levels, pressure, throttle, and speed) and view diagnostic reports without having to remove the engine from the hull. Moreover, the tank's driver could be alerted to critical engine problems with an onboard display.

"By using TEDANN, the Army could expect to extend the life of the M1 Abrams tank fleet, lengthen the time between costly overhauls by 25%, and avoid hundreds of millions of dollars in maintenance costs over 30 years," maintains Frank Greitzer, program manager for Pacific Northwest.