Troubleshooting machines
Rotating machinery (such as motors, turbines, transmissions, gears, etc) are ubiquitous in factories all over the world.
Such mechanisms are often complex, and it may be difficult to identify the root causes of their failures from symptoms measured by sensors.
To solve this diagnosis issue, a probabilistic “twin” of the rotating machine can be built, allowing to model the system’s behavior in a generative way, which can be inverted through classical Bayesian inference to trace back to the most likely causes of failuresfrom the sensor readings.
As a benefit, the maintenance time is significantly reduced as operators have better insights on what is wrong with the machine, yielding reduced operating costs.