Water-turbine generator set multi-working-condition abnormity monitoring method and device based on sound recognition

The invention provides a water-turbine generator set multi-working-condition abnormity monitoring method and device based on sound recognition. The method comprises the steps that sound data of a water turbine under different working conditions are collected; performing mode extraction on the sound data under different working conditions to obtain mode sequences under different working conditions to establish a database; training the built time-frequency domain fused self-encoding feature extraction network by fusing the sound data in the database to obtain a trained self-encoding feature extraction network; constructing a twin feature extraction network based on the trunk of the trained self-encoding feature extraction network; collecting real-time sound signal data under a current working condition and a mode sequence under a corresponding working condition in a database, and inputting the real-time sound signal data and the mode sequence into a twin feature extraction network to obtain the similarity of tw