Feature selection for damage degree classification of planetary gearboxes using support vector machine

Abstract

Feature selection is an effective way of improving classification, reducing feature dimension, and speeding up computation. This work studies a reported support vector machine (SVM) based method of feature selection. Our results reveal discrepancies in both its feature ranking and feature selection schemes. Modifications are thus made on which our SVM-based method of feature selection is proposed. Using the weighting fusion technique and the one-against-all approach, our binary model has been extensively updated for multi-class classification problems. Three benchmark datasets are employed to demonstrate the performance of the proposed method. The multi-class model of the proposed method is also used for feature selection in planetary gear damage degree classification. The results of all datasets exhibit the consistently effective classification made possible by the proposed method.

Gear pitting severity level identification using binary segmentation methodology

“…Qu et al seeded the pitting fault in the gear of different severity levels (initial, medium, and severe pitting). A total of 134 time domain and frequency domain‐based features were extracted from the raw vibration signals.…”

“…The classification approaches such as SVM, KNN, LDA, GRA, and ORM are based on a supervised learning framework and requires both input (features extracted from the raw vibration data) and target variable (exact health stage [initial, medium, or severe pitting stage] of the gear corresponding to the particular feature value) for model training. In all the past reported work, the fault severity classification approaches are applied to the seeded fault, and hence, exact state change point is known in advance. However, in practice, the gear tooth is subjected to a natural pitting progression, and exact state change points of the gear are not known a priori.…”

Fault features analysis of a compound planetary gear set with damaged planet gears

“…Many important research topics have been proposed, including mathematical signal models, [1][2][3][4][5] model-based simulation, [6][7][8][9][10][11][12][13][14][15][16][17][18] advanced signal processing methods, 19,20 and fault features extraction. [21][22][23][24][25] Most studies are focused on fault-induced impulses detection in the vibrations and modulation sidebands analysis in the spectra. The contacts of the damaged tooth area with the mating gears will cause impulses in the vibration.…”

“…Furthermore, they 15 proposed a windowing and mapping strategy to detect the ring gear side tooth crack of the planet gear. Qu et al 24 used the SVM-based feature selection method to assess the pitting degrees of the planet gear of a planetary gear set. Liu et al 25 combined the kernel feature selection method with the kernel Fisher discriminant analysis (KFDA) to assess the pitting levels of the planet gear of a planetary gear set.…”

Evaluation of the spur gear condition using extended frequency range

“…Diagnostic parameters used in time domain are presented in Fig. 1, more information about these parameters can be found in [6,10,14,16,19,21,24,25,26,38,39]. Some of them can be also used in frequency and in time-frequency domains.…”