Rolling bearing fault diagnosis method based on improved multi-scale fuzzy entropy

Abstract

The invention relates to a rolling bearing fault diagnosis method based on an improved multi-scale fuzzy entropy. A vibration signal of a rolling bearing is collected; an improved multi-scale fuzzy entropy of the vibration signal is calculated; the improved fuzzy entropies of the first eight scales are used as bearing fault feature vectors; the fault feature vectors are classified into a training set and a testing set; with the training set, a support vector machine is trained and the testing set is predicted by using the trained model; and according to a prediction result, a working state and a fault type of the rolling bearing are identified. On the basis of improvement of the fuzzy entropy algorithm, a total mean value is used for replacing a local mean value in the traditional fuzzy entropy calculation and improved fuzzy entropies under different scales are calculated. Because of the improved multi-scale fuzzy entropy, the signal features can be reflected comprehensively and thus the operating state of the bearing can be evaluated accurately. Therefore, the diversified bearing state information can be extracted; and the recognition rate during the fault mode recognition process is improved.

Description

technical field [0001] The invention relates to a fault diagnosis technology, in particular to a rolling bearing fault diagnosis method based on improved multi-scale fuzzy entropy. Background technique [0002] Rolling bearings are one of the key components in rotating machinery, and their operating status often determines the performance of the entire machine. Therefore, the fault diagnosis of rolling bearings is of great significance. Among various bearing fault diagnosis methods, the diagnosis based on vibration signal is one of the most commonly used and most effective methods. However, rolling bearings will inevitably be affected by nonlinear factors such as friction, clearance and nonlinear stiffness during operation, and the collected vibration signals often present strong nonlinear and unsteady characteristics. Therefore, the traditional linear system-based time-domain and time-frequency domain signal analysis methods are difficult to accurately extract bearing fau...

AI Technical Summary

Problems solved by technology

[0003] However, fuzzy entropy subtracts a local mean when constructing the vector needed for calculation, which makes the overall trend of the signal ignored when calculating entropy

Images