Dredge pump wear degree identification method based on deep learning acoustic scene classification

The invention discloses a dredge pump wear degree identification method based on deep learning acoustic scene classification. As the service time of the dredge pump equipment increases, friction, abrasion or other faults may occur to internal parts, resulting in the emission of sound signals different from normal operation states. In order to ensure the dredging efficiency, the dredge pump needs to be checked and maintained regularly. The sound signals of the deep learning model dredge pump equipment are used for classification modeling so as to distinguish the normal state from the wear state. A large number of sound samples in normal and wear states are used through the training model, and the model can learn sound features in different states and perform accurate classification and identification. Thus, once the abrasion sound is detected, maintenance or replacement measures can be taken in time to avoid further damage to equipment. 本发明公开了一种基于深度学习声学场景分类的泥泵磨损程度识别方法。随着泥泵装备的使用时间增加，内部部件可能会出现摩擦、磨损或其他故障，导致发出不同于正