Abstract:The rotor system is the core equipment of industrial rotating machinery, and ensuring its safety is an essential basis for improving the productivity of the equipment. As a critical monitoring quantity reflecting the operating status of the rotor system, identification models based on axis orbits are effective means for detecting equipment faults. However, most of the existing axis orbit identification models belong to the category of image recognition, and these methods have defects such as unclear physical meaning of features and weak generalization performance. Therefore, the paper returns to the essence of axis orbits and proposes a rotor axis orbit recognition method based on multivariate swing signals, feature extraction and pattern recognition. Firstly, the mutually perpendicular swing signals of the rotor are obtained based on eddy current sensors. Secondly, we propose a feature extraction tool for extracting the multivariate signals named enhanced hierarchical multivariate fuzzy entropy (EHM v FE), a nonlinear dynamics metric based on the enhanced hierarchical decomposition method. Next, the features of axis orbits are extracted by the EHM v FE. Finally, some of the extracted features are input into an extreme learning machine (ELM) for model training, and the effectiveness of the method is verified with the remaining samples. We apply the proposed method to the rotor axis orbit identification case, and the results show that its recognition rate is 98.963%. In comparison experiments with recognition models based on nonlinear dynamics indicators, multivariate signal processing methods, traditional image feature extraction methods, and popular deep learning models, the proposed model shows substantial advantages, verifying the reasonableness and superiority of the proposed method. This study provides a new idea for rotor shaft fault diagnosis, which has significant reference value for promoting the development of intelligent operation and maintenance of industrial equipment.

Application wavelet modulus maxima to the identification of shaft center orbit

IDENTIFICATION OF THE SHAFT ORBIT FOR ROTATING MACHINES USING WAVELET MODULUS MAXIMA

Singular Signal Feature Extraction and Identification Based on Empirical Orthogonal Function for Module Maxim and Phase Matrix of Wavelet

Identification The Orbits Of Shaft Centerline Based On Wavelet Transforms And Ar-Model

Morphological haar wavelet-based purification of rotor center's orbit

Bearing Fault Detection Via Wavelet Packet Transform and Rough Set Theory

Early Weak Fault Signal Enhancement and Recognition Method of Rudder Paddle Bearings Based on Parameter Adaptive Stochastic Resonance

On Identifying Rotating Machinery Orbits Of Shaft Centerline Based On Neural Networks

Novel Bearing Fault Diagnosis Model Integrated with Dual-Tree Complex Wavelet Transform, Permutation Entropy and Optimized FCM

Feature Extraction of Weak-Bearing Faults Based on Laplace Wavelet and Orthogonal Matching Pursuit

Combination of Discrete Cosine Transform with Neural Network in Fault Diagnosis for Rotating Machinery

Bearing Fault Diagnosis Using Wavelet Domain Operator-Based Signal Separation

Simulation Research on Exploiting Electrostatic Potential Difference Sensing for Shaft Centerline Orbit Reconstruction.

Application of K-Nearest Neighbor Algorithm Based on Orthogonal Wavelet Feature Extraction in Fault Diagnosis

Maximal-overlap Adaptive Multiwavelet for Detecting Transient Vibration Responses from Dynastic Signal of Rotating Machineries

Customized Multiwavelets for Planetary Gearbox Fault Detection Based on Vibration Sensor Signals.

Wavelet-Enhanced Time-Frequency Analysis Method for Bearing Fault Detection of Rotating Machinery

Fault feature extraction and diagnosis of rolling bearings based on wavelet thresholding denoising with CEEMDAN energy entropy and PSO-LSSVM

An Identification Method for Rotor Axis Orbits based on Enhanced Hierarchical Multivariate Fuzzy Entropy and Extreme Learning Machine

The Detection of Motor Bearing Fault with Maximal Overlap Discrete Wavelet Packet Transform and Teager Energy Adaptive Spectral Kurtosis

LW-BPNN: A Novel Feature Extraction Method for Rolling Bearing Fault Diagnosis