A novel method for fault diagnosis in rolling bearings based on bispectrum signals and combined feature extraction algorithms
Autor: | Zohreh Hashempour, Hamed Agahi, Azar Mahmoodzadeh |
---|---|
Rok vydání: | 2021 |
Předmět: |
Bearing (mechanical)
Artificial neural network business.industry Computer science Gaussian Scale-invariant feature transform Pattern recognition law.invention symbols.namesake Histogram of oriented gradients law Signal Processing Genetic algorithm symbols Artificial intelligence Electrical and Electronic Engineering business Bispectrum Induction motor |
Zdroj: | Signal, Image and Video Processing. 16:1043-1051 |
ISSN: | 1863-1711 1863-1703 |
DOI: | 10.1007/s11760-021-02053-7 |
Popis: | Rolling bearings are vital components in the induction motors. Rolling bearings are encountered to extremely tensions, and their faults can cause serious damages to induction motors; therefore, it is necessary to diagnose their faults as soon as possible. Although the instantaneous performance of the rolling bearings can be recorded using sensors, due to unexpected, nonlinear and unstable vibrations of motors, these recorded signals are mixed with linear and nonlinear noises. This research introduces a novel algorithm for this problem by employing multiple novel methods. The Gaussian noises are removed by wavelet decomposition, and nonlinear noises are suppressed by the bispectrum method. The latter method maps a 1-D signal into a 2-D signal. Therefore, for extracting features, it is possible to use feature description algorithms developed for images. In this paper, the histogram of oriented gradients (HOG), the scale-invariant feature transform (SIFT) and the speeded up robust features (SURF) descriptors are used for this purpose. In addition, for each descriptor, the best features are selected using the genetic algorithm (GA). Finally, the selected features are combined together and applied to the generalized regression neural network (GRNN). The performance of the proposed algorithm is assessed on the standard bearing dataset of the Case Western Reserve University. |
Databáze: | OpenAIRE |
Externí odkaz: |