Beschreibung:
Abstract At present the most widely used method in mechanical fault diagnosis is feature extraction based on vibration signal analysis, but the test results are not accurate. Fault diagnosis method based on acoustic measurement can overcome this shortcoming, although traditional acoustic measurement based on single channel is susceptible to interference and it is difficult to obtain effective signals under strong background noise or complex testing environment. Directional noise measurement technology based on acoustic array and beamforming method can effectively suppress background noise. In this paper, the 3D constant beam width beam forming array is used to extract the broadband radiated noise from the monitored equipment and the sensitive frequency bands corresponding to different fault stages of bearings are preprocessed to obtain the corresponding eigenvectors. BP Neural Network is used to identify the fault of the tested bearings. The results show that the 3D microphone array based on constant beam width beam forming can effectively suppress the interference of environmental noise to ensure the validity of monitoring data and improve the reliability of fault diagnosis.