Data mining and probabilistic risk assessment will help improve safety, experts say.

Laurie Sullivan, Contributor

January 30, 2006

2 Min Read

Three years after the space shuttle Columbia disintegrated above Texas on reentry Feb. 1, scientists are looking more closely at advanced predictive analysis software to ensure a similar disaster doesn’t occur again.

Science Applications International Corp. on Monday said it is developing Fourier analysis to integrate with Insightful Corp.'s data mining analytics tool Insightful Miner 7. "Since Challenger, and even Columbia, we now have more advanced safety methods and the ability to look for different kinds of data," said Travis Moebes, senior scientist for Science Applications. "Two analysis tools we use today that weren't around before are data mining and probabilistic risk assessment."

Moebes's technical team, which provides the National Aeronautics and Space Administration with IT and analysis support, is responsible for data management. It mines and analyzes data to support safety for NASA’s Shuttle, Space Station and other spaceship programs using science and statistical methods.

The largest customers of data mining tools have been banks and pharmaceutical companies. Retail stores have been adopting the technology as well. In the business world, "you might predict who is not a good customer to issue a credit card, but not for the reasons you might suspect," Moebes said. "In the scientific and space world, you're trying to save lives."

And much can go wrong. The Space Shuttle, for example, is built with more than 10,000 "major parts," not including nuts and bolts. Moebes said the data mining trend in space travel will open. Virgin Galactic, Richard Branson's commercial endeavor into space, will require a similar "safety program and these tools could support the analysis."

The idea is to simulate a condition to determine the outcome. To identify a potential problem, the software plots "what-if" scenarios to determine the outcome. If a valve sticks it's important to know the conditions surrounding the problem prior, as well as at the time the event occurs.

A power surge or change in oxygen saturation could cause a problem, for example. "Sometimes it's not apparent why a change in atmosphere would cause a part to stick," Moebes said. "You look at the conditions and surrounding problems around during the time it happened, and then the engineer decides if it connects or if it's a statistical coincidence."

By deploying Insightful Miner 7 along with in-house text mining applications, Science Applications is able to reduce the amount of time it takes to analyze NASA Corrective Action Reports. One engineer can now identify problems in three to four hours, where it once took eight engineers six weeks.

About the Author(s)

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights