Bearing Fault Diagnosis Based on Subband Time-Frequency Texture Tensor

Abstract

The texture feature tensor established from a subband time-frequency image (TFI) was extracted and used to identify the fault states of a rolling bearing. The TFI of adaptive optimal-kernel distribution was optimally partitioned into TFI blocks based on the minimum frequency band entropy. The texture features were extracted from the co-occurrence matrix of each TFI block. Based on the order of the segmented frequency bands, the texture feature tensor was constructed using the multidimensional feature vectors from all the blocks; this preserved the inherent characteristic of the TFI structure and avoided the information loss caused by vectorizing multidimensional features. The linear support higher order tensor machine based on the feature tensor was applied to identify the fault states of the rolling bearing. INDEX TERMS Texture feature tensor, frequency band entropy, linear support higher-order tensor machine, bearing fault intelligent diagnosis.

Fractional Synchrosqueezing Transformation and its Application in the Estimation of the Instantaneous Frequency of a Rolling Bearing

“…To verify the noise robustness of the FRSST, different SNRs are added to the signal in (41), and the signals with noise are obtained as 21…”

“…These methods usually require the installation of an additional key-phase device to measure the actual speed of the bearings, but it is difficult to implement when the installation of the device is inconvenient. The other category involves methods with no key-phase device focused on the IF estimation that is mainly based on the phase demodulation [15][16][17][18] and time-frequency representation [19][20][21][22][23]. The former type of method demodulates the harmonic signal extracted from the rolling bearing vibration signals to obtain instantaneous phase information, but it can only obtain this information when the harmonics constantly exist in the rolling bearing vibration signal and have a sufficiently high energy level.…”

Multi-Band Frequency Window for Time-Frequency Fault Diagnosis of Induction Machines

“…To extend MCSA to the fault diagnosis of IM working in transient regime, advanced time-frequency (TF) transforms of the current are needed, so that the transient fault signatures can be identified in a joint TF domain. These transforms can be linear, such as the short-time Fourier transform (STFT) [28][29][30][31], the short-frequency Fourier transform (SFFT) [31,32] and the wavelet transform (WT) [33], or quadratic, such as the Wigner-Ville distribution (WVD) [34] or the ambiguity function [8]. Quadratic TF transforms can achieve optimal resolution for mono-component chirp signals but, in case of multi-component ones, they produce cross-terms artifacts that pollute the TF representation of the current, making it difficult the correct identification of the fault harmonics.…”

“…In this way, the Gaussian window has a duration-bandwidth product (7) that reaches the minimum value (i.e., the highest concentration in the joint TF plane) that can be achieved under the uncertainty principle (8).…”

“…A way to reduce these risks is the continuous monitoring of the machine's condition, in order to detect the presence of a fault at an early stage, when corrective measures can be implemented or maintenance works scheduled. Diverse quantities have been proposed in the technical literature for implementing condition based maintenance systems (CBMS) [1,2], such as the analysis of the stator currents [3][4][5][6][7], machine vibrations [8][9][10], fluxes [11,12], thermal images [13] or acoustic signals [14,15]. These techniques have been applied to detect different types of faults not only of the IM, such as stator inter-turn short circuits [11,16], broken bars [17,18], rotor asymmetries [19], eccentricity [20], bearing faults [6,21], but also of the inverter drive [3] or the mechanical coupling to the load, as gearboxes and pulleys [3,22].…”