Bearing fault diagnosis method based on particle swarm optimization with stacked noise reduction self-encoding network

Abstract

The invention discloses a method for diagnosing bearing faults in stacked noise-reducing self-encoding networks based on particle swarm optimization, and proposes an improved method for diagnosing faults in SDAE bearings in stacked-noise-reducing self-encoding networks. The number of nodes in the hidden layer of the network, sparse parameters, and the proportion of input data are randomly set to zero for adaptive selection to determine the SDAE network structure, based on which the high-level feature representation of the fault state is obtained, and input to the Soft-max classifier for fault classification and identification ; The present invention not only has better feature learning ability, but also has more robustness compared with the features learned by ordinary sparse autoencoders, and optimizes the hyperparameters of the noise reduction autoencoder deep network structure through the particle swarm optimization algorithm, and constructs The SDAE diagnosis model with multiple hidden layers ultimately improves the accuracy of fault classification.

Description

technical field [0001] The invention belongs to the technical field of mechanical manufacturing and relates to a mechanical fault diagnosis technology, in particular to a particle swarm optimization-based stacking noise reduction self-encoding network bearing fault diagnosis method. Background technique [0002] As a common component of rotating machinery, rolling bearings may cause major economic losses once they fail during work. Therefore, effective diagnosis and treatment of rolling bearing failures is of great significance to ensure the normal operation of the machine. [0003] The fault diagnosis method based on artificial intelligence has been widely used in the fault diagnosis of rotating machinery and achieved good results. At this stage, the fault diagnosis of rolling bearings mostly judges its running state through the detection and analysis of various state parameters, and determines the fault location and wear degree. Generally, bearing fault diagnosis can be c...

AI Technical Summary

Problems solved by technology

At present, most of its hyperparameters are determined by empirically enumerating a variety of hyperparameter combinations to obtain a better set of hyperparameters. For fault diagnosis problems, especially for fault classification problems in different fields, the generalization performance is weak.

Images