Equipment for detecting non-stational vibration, methods for detecting non-stational vibration, methods for determining threshold values, and recording devices

Abstract

The invention relates to an artificial intelligence trained to detect the non-stationary nature of an object, which performs this function accurately even in the presence of ambient sound. The static vibration characteristic data is generated from static vibration data, which is data about static vibration including vibrations generated in a static state from the object, where the detection of non-stationary nature is performed based on sound. Static vibration characteristic data is data about the characteristics of static vibration identified by static vibration data, which is fed to an automatic encoder to cause the automatic encoder to produce estimated static vibration characteristic data. A loss function between the static vibration characteristic data and the estimated static vibration characteristic data is generated, and the automatic encoder is trained to minimize the difference between the two types. By repeating the above processing procedure, the trained autoencoder receives the received message from the autoencoder.