Disney patents AI system to detect ride failures via sound

Disney has filed a patent for an artificial intelligence system designed to detect equipment failures on theme park rides by monitoring acoustic signatures. The technology aims to identify mechanical issues before they result in visible malfunctions or unexpected ride shutdowns.

The system utilizes scientific microphones equipped with cylindrical wind filters to isolate machine sounds from ambient park noise. To ensure efficiency, sensors trigger recordings only when specific equipment, such as a track switch or actuator, is in motion.

The process employs two distinct machine learning models. The first model is trained to filter out human noise, such as guest voices and screaming, from the mechanical audio. The second model, an autoencoder, compares real-time recordings against a baseline of normal operating sounds. By tracking reconstruction errors, the system can statistically identify deviations that indicate potential mechanical problems.