Rolling bearing fault diagnosis method based on double sparse dictionary sparse representation

Abstract

The invention belongs to the technical field of rolling bearing fault diagnosis. The invention discloses a rolling bearing fault diagnosis method based on double sparse dictionary sparse representation, and the method comprises the steps: S1, employing a double sparse dictionary learning algorithm to train a rolling bearing vibration signal, and obtaining a double sparse dictionary; S2, obtainingdecomposition coefficients of the modeling rolling bearing vibration signals of different fault types under the double sparse dictionaries and taking the decomposition coefficients as feature vectors;S3, inputting the feature vector obtained in the step S2 into a deep belief network for training and learning to obtain a rolling bearing vibration signal fault diagnosis model; and S4, inputting theto-be-detected rolling bearing vibration signal containing the fault information into the fault diagnosis model for fault identification, thereby finishing fault diagnosis. By the adoption of the rolling bearing vibration signal fault diagnosis method, higher diagnosis precision and accuracy stability can be obtained, the training and testing time of the deep belief network can be greatly shortened, and the fault diagnosis efficiency is improved.

Description

technical field [0001] The invention belongs to the technical field of rolling bearing fault diagnosis, and in particular relates to a rolling bearing fault diagnosis method based on a double-sparse dictionary sparse representation combined with a deep belief network. Background technique [0002] Rolling bearings are the most widely used key components in rotating machinery, and their operating status has a direct impact on the overall performance of the entire machine. Once a failure occurs, downtime for maintenance or even major accidents will be inevitable. Mechanical vibration signals can best reflect the state information of rolling bearings, so fault diagnosis of vibration signals in the working state of rolling bearings is an effective method to reduce downtime and maintenance costs and ensure production safety. [0003] However, since the mechanical vibration signal itself inevitably contains some noise information, which is non-stationary and nonlinear, it is not o...

AI Technical Summary

Problems solved by technology

The analysis dictionary is a highly structured mathematical model, which has the characteristics of fast solution, but at the same time shows the disadvantage of poor adaptability to the signal. Such dictionaries include wavelet dictionary, DCT dictionary, curve wave dictionary, etc.; learning dictionary is adopted The machine learning algorithm obtains an over-complete dictionary from a set of sample signals, so it has good adaptability, but due to unstructured, it is difficult to solve. Such dictionaries include PCA, MOD and K-SVD

[0005] Therefore, due to the above-mentioned problems in the processing of mechanical vibration signals in the process of rolling bearing fault diagnosis using the conventional sparse representation of the analysis dictionary or learning dictionary, the efficiency of dictionary learning is low, which in turn affects the accuracy and accuracy of the entire rolling bearing fault diagnosis.

Images