Fault state recognition method for port machine wheel bearing

Abstract

The invention discloses a fault state recognition method for a port machine wheel bearing. The fault state recognition method comprises the steps that collected vibration signal data of the port machine wheel bearing in different fault states are evenly divided into a plurality of samples, and classified label matrixes are constructed and formed; time domain characteristic extraction, frequency domain characteristic extraction and time-frequency domain characteristic extraction are conducted on all the samples to obtain the characteristic number, and a high-dimensional original characteristicdata set matrix with the sample total number as the line number and the characteristic number as the column number is established; a Pearson correlation coefficient and residual analysis combined method is adopted to conduct correlation analysis on obtained characteristic parameters, distinguishing characteristics are extracted to reduce the dimension of the high-dimensional original characteristic data set matrix, and thus a sample characteristic matrix is obtained; and the dimension-reduced sample characteristic matrix and the corresponding label matrix are led into a deep belief network algorithm for classified training, and thus a bearing fault state classifying model is obtained. The bearing fault can be detected accurately and effectively, and the defects of a traditional method in bearing fault recognition are overcome.

Description

technical field [0001] The invention relates to the technical field of fault monitoring and diagnosis of equipment systems, in particular to a method for identifying fault states of wheel bearings of port machinery based on correlation coefficients and deep learning. Background technique [0002] Port machinery trolley bearing is a rolling bearing with large load capacity, low working speed and frequent work, which bears a large load during operation. Once the bearing fails, it will seriously affect the production. At the same time, the corresponding maintenance and replacement of the bearing will take a long time, which will lead to large economic losses. Therefore, it is necessary to monitor the bearing fault status to improve production efficiency. The traditional bearing fault detection technology mainly includes ferrography diagnosis technology and temperature diagnosis technology, etc., but the iron shop diagnosis technology is only suitable for the bearing fault diag...

AI Technical Summary

Problems solved by technology

The traditional bearing fault detection technology mainly includes ferrography diagnosis technology and temperature diagnosis technology, etc., but the iron shop diagnosis technology is only suitable for the bearing fault diagnosis of lubricating oil lubrication, and it is not generalizable; the temperature diagnosis technology is good for judging bearing burns, but Simple routine diagnostics only for bearings in machines

Images