NASA - National Aeronautics and Space Administration

Finding defects in rotating machinery will be quicker and easier, thanks to two new NASA computer programs and an award-winning technical paper that describes them.

Two NASA computer scientists who wrote about software that can find faults in gearing, received an award for their technical paper from the Society of Automotive Engineers (SAE, International), Warrendale, Pa. SAE represents engineers in mobility-related professions.

"The significance of this technology is that from sensory data, it can predict if, and at what location, a given moving part of a mechanical system may fail during operation," said Upender Kaul of NASA Ames Research Center, Moffett Field, Calif., one of the two authors and principal investigator of the project.

Image left: The new software will be able to detect defects in a complicated machine with multiple moving parts. Image credit NASA.

SAE gave computer scientists Kaul and Nikunj Oza, also of NASA Ames, the Arch T. Colwell Merit Award at the SAE 2007 World Congress in Detroit in April 2007. Kaul and Oza co-authored the research paper, "Machine Learning for Detecting and Locating Damage in a Rotating Gear." The paper explains how the software analyzes stress and other data from normal gears, and then can detect damaged gear teeth.

The article not only describes a method to detect and locate damage in a rotating gear using the two programs, but also notes that the detection process is not limited to gearing. One program involves modeling and simulation software that can simulate machine subsystems. The second program involves software that can predict how well a machine will operate.

"Based on this technology, we can create damage detectors and locators that can be more robust," Oza said.

Kaul developed the simulation and machine-learning software. Oza fine-tuned the machine-learning software to enable the computer programs to detect faults more efficiently. The work reflected in the technical paper was primarily funded by the NASA’s Computing, Information and Communications Technology (CICT) program managed by Eugene Tu of NASA Ames Research Center.

Kaul has more than 27 years of research and development experience in the field of computational sciences and has published more than 60 technical documents, including papers in peer-reviewed conferences and journals. Oza wrote more than 20 papers in peer-reviewed conferences and journals.

Image left: The software can find hard to detect defect. Photo credit: NASA.

According to SAE, the society creates and manages more aerospace and ground vehicle standards than any other entity in the world. "Today, as SAE approaches its 100th anniversary, we represent the collective wisdom of nearly 85,000 engineers, technical professionals, academics and governmental representatives in 97 countries around the globe," the society Web site observes.
John Bluck
NASA Ames Research Center, Moffett Field, Calif.
Phone: 650-604-5026 email: jbluck@mail.arc.nasa.gov