Method and device for fault diagnosis of rotating parts based on improved cnn and relational module

Abstract

The invention discloses a fault diagnosis method for rotating parts based on improved CNN and relational modules, constructs a fault diagnosis metadata set, and divides it into a training set and a test set according to fault categories; performs fast Fourier transform on the original data set samples; uses multiple Three strategies of scale convolution kernel, random pooling, and hole convolution are used to establish a convolutional neural network diagnostic model consisting of an extraction module, a fusion module, and a relational module; the model is trained using the training set using the meta-learning method; the test set is used to pair The trained model performs small-sample multi-classification rotating component fault diagnosis. The present invention can self-adaptively train the measurement standard of the distance between samples, and use the characteristics of meta-learning to quickly diagnose new faults with only one labeled sample, thereby solving the problem of traditional methods relying on large amounts of data and long-term training. It effectively solves the problem of cross-domain diagnosis of new faults under the condition of small samples.

Description

technical field [0001] The invention relates to the field of fault diagnosis of mechanical equipment, in particular to a fault diagnosis method and a fault diagnosis device for a rotating component based on an improved CNN and a relational module, especially a rotating mechanical component. Background technique [0002] Due to the rapid development of modern industry, rotating machinery is widely used in navigation, machinery, chemical industry and other fields. If the key equipment fails, it will cause unpredictable losses. Therefore, it is of great significance for the fault diagnosis of rotating machinery. Existing deep learning methods are widely used in the field of fault diagnosis of rotating parts of rotating machinery and have achieved good results. However, they still face the following two problems. (1) Most mechanical equipment operates in a normal state, and the fault data is relatively large. It is difficult to obtain, but training a deep learning model require...

AI Technical Summary

Problems solved by technology

Existing deep learning methods are widely used in the field of fault diagnosis of rotating parts of rotating machinery and have achieved good results. However, they still face the following two problems. (1) Most mechanical equipment operates in a normal state, and the fault data is relatively large. It is difficult to obtain, but training a deep learning model requires a large amount of labeled data. Secondly, rotating machinery often operates under variable working conditions, which will lead to differences between samples. Therefore, using a small number of samples to retrain the model has certain advantages. challenging

(2) The occurrence of new faults in the process of mechanical operation is unknown, and the judgment of new faults requires experts to diagnose and identify them. Therefore, the samples of faults are limited

Every time a new fault occurs, the model requires a large amount of data for training, which undoubtedly consumes manpower and time

Images