A Bearing Fault Diagnosis Method for Automatic Encoding Machine with Adaptive Parameter Adjustment

Abstract

The invention discloses a bearing fault diagnosis method based on an adaptive parameter adjustment automatic coding machine. The method comprises the following basic steps: 1) bearing vibration signals are sampled; 2) bearing signals are preprocessed; 3) the node number and the structure of a depth network cost function are determined; 4) adaptive parameter adjustment is carried out; and 5) faultdividing is carried out. The method firstly carries out noise reduction processing on the current data, dimension reduction processing is carried out while noise reduction is carried out, and the depth network is entered through normalized clean data for training; through the characteristics of a sparse automatic encoder, sparsity restriction is carried out on hidden layer neurons in an edge noisereduction and moving encoder; an Ada-grad learning strategy is combined to continuously adjust the parameter of the current learning rate to enable the parameter to be optimal, and a fast-convergenceand high-accuracy classification effect is thus achieved. Finally, through comparison with the traditional automatic coding machine in bearing fault classification, the characteristics of strong effectiveness and robustness of the method of the invention can thus be verified.

Description

technical field [0001] The invention relates to a fault feature classification method of an adaptive parameter sparse edge noise reduction automatic encoder, which belongs to the field of rolling bearing fault diagnosis. Background technique [0002] In mechanical equipment, the abnormal state of rolling bearings will cause abnormal noise or vibration of the equipment and even cause damage to the equipment. Some data show that rolling bearing failures account for about 30% of rotating machinery failures. At the same time, if the machine is shut down for inspection and maintenance when a failure occurs, it will greatly affect the production. For some large-scale production, the machine downtime means a lot of economic losses. Therefore, it is of far-reaching practical significance to accurately identify the early state of rolling bearings before failure. [0003] Although fault diagnosis technology has been greatly developed in recent years and has received more and more at...

AI Technical Summary

Problems solved by technology

However, due to the random selection of the number of nodes in the hidden layer and the preset learning rate, the model cannot achieve good results in actual operation.

Images