A Fault Diagnosis Method with Noisy Labels Based on Recurrent Neural Networks

Abstract

The invention relates to a rotary machine fault diagnosis method, in particular to a fault diagnosis method with a noise label based on a recurrent neural network. According to the invention, the problem of low diagnosis accuracy of the existing rotary machine fault diagnosis method is solved. The fault diagnosis method with a noise label based on a recurrent neural network comprises the followingsteps: 1, carrying out segmentation normalization of vibration signals, and forming a data set; 2, adding a noise label of a data sample into the data set to form a noise label data set; 3, adjustingthe parameters of the whole network by using an optimization algorithm to minimize an adaptive correction loss function, thereby completing the training of the recurrent neural network; and 4, inputting a test data set into the trained recurrent neural network, and outputting a fault diagnosis result. The method is suitable for fault diagnosis of a rotary machine.

Description

technical field [0001] The invention relates to a fault diagnosis method for rotating machinery, in particular to a fault diagnosis method with noise labels based on a cyclic neural network. Background technique [0002] Rotating machinery is the core component of mechanical equipment, and its operating status is closely related to the safety and reliability of mechanical equipment. Due to the harsh working environment, rather complex structure and highly automated operation, rotating machinery is prone to various failures. Any failure that cannot be detected may quickly trigger a chain reaction, damage other components, generate huge economic losses, and even threaten personal safety. Therefore, there is a great need to develop and improve fault diagnosis methods for rotating machinery. However, the practice shows that the existing rotating machinery fault diagnosis methods have the problem of low diagnosis accuracy because they ignore a large number of noise labels in th...

AI Technical Summary

Problems solved by technology

[0003] In order to solve the problem of low diagnostic accuracy of existing rotating machinery fault diagnosis methods, the present invention provides a fault diagnosis method with noise labels based on cyclic neural network

Images