A neural network originally trained to distinguish real gravitational waves from noise in a gravitational wave detector recognizes bearing wear from vibrations without retraining. False alarms, or glitches, match the time structure exactly of the short, frequent impacts that occur when metal degrades inside machinery. The net picks up these 'clicks' much like a doctor hears heart murmurs through a stethoscope.
The method doesn't react to electrical noise: that's a different physical beast. It relies on a spectroscopic approach: the program analyzes the frequency and time pattern of the signal, not its source. This way, without any retraining, you can monitor the health of all sorts of equipment, from wind turbines to factory robots, predicting breakdowns long before disaster strikes.
For decades, false alarms frustrated gravitational wave hunters — now they're a boon for machine diagnostics. Noise became the instrument.
In Carl Sagan's novel 'Contact,' a message from space delivered blueprints for a machine. Here, cosmic noise becomes a stethoscope for earthly machinery.