Vibration image-driven rolling bearing intelligent fault diagnosis method

Abstract

The invention discloses a vibration image-driven rolling bearing intelligent fault diagnosis method, which comprises the following steps of: firstly, selecting a time domain characteristic index with the maximum discrimination rate for rolling bearing data, calculating the angle self-adaption rate of each layer of IMF according to the time domain characteristic index, and then performing EMD-AADPC vibration image conversion on vibration data, generating EMD-AADPC vibration image samples, and dividing training, verification and test data sets; and building a CNN classification model, performing training by using a training set sample, performing model hyper-parameter optimization adjustment according to a verification set result to obtain an optimal model training parameter, and performing model diagnosis performance evaluation by using a test data set. According to the invention, fault diagnosis of the rolling bearing is realized, and the diagnosis mode becomes more intelligent and efficient.

Description

technical field [0001] The present invention relates to the field of mechanical fault diagnosis, in particular to the direction of rolling bearing fault diagnosis, in particular to an EMD-AADPC (Empirical Mode Decomposition-Adaptive Angle Distribution PolarCoordinate, empirical mode decomposition-adaptive angle distribution polar coordinates) vibration image and CNN ( ConvolutionalNeural Network, Convolutional Neural Network) Rolling Bearing Fault Diagnosis Method Background technique [0002] The safe operation of mechanical equipment is an important guarantee for modern industrial production. Bearings are an indispensable part of mechanical equipment, and the fault diagnosis technology of rolling bearings has received great attention. Rolling bearings are generally composed of inner rings, outer rings, rolling elements and cages. After a long period of operation, various faults are prone to occur. Therefore, the rapid and accurate identification of rolling bearing faults i...

AI Technical Summary

Problems solved by technology

CNN is a feed-forward neural network, which has a good ability to extract data feature representation, and shows its superiority when processing images, so there has been a diagnostic method that uses image data to drive classification models; existing vibration image-driven In the method, the characteristics of the vibration image samples between different state data are not clear, the characteristic information contained is single, and the characteristics of the samples cannot be adaptively enhanced.

Images