Artificial intelligence (AI) holds great promise for helping humans make informed decisions and boost business efficiency, but so far, many applications for the supply chain are more theoretical than practical.
Now a team of researchers from the University of Pittsburgh has come up with a way to use the technology to literally keep the trains running on time. The engineers taught an AI program on a supercomputer at the Pittsburgh Supercomputing Center (PSC) to predict when steel railroad rails might buckle or break, before an accident happens.
The team used PSC’s massive Bridges-2 supercomputer to simulate how the temperature-induced expansion and contraction of rail segments affects sound moving through them, and then trained the AI to virtually “listen” to those signals and flag potential problems in a given stretch of track.
According to the team, that approach is more efficient than the industry’s current method of monitoring rails, which involves closing down a section of railway, cutting out samples of the metal for testing, and repairing it afterward.
The researchers are now working on expanding their dataset, with a goal of simulating another 4,000 to 4,500 additional cases. They then hope to install the technology in either stationary, trackside “wayside detectors” or in mobile “in-motion detectors” mounted on inspection vehicles that visit multiple sites.
Source link
