Accesso libero

GGLCM: A Real-time Early Anomaly Detection Method for Mechanical Vibration Data with Missing Labels

Hu Yu

Hu Yu

College of Mechanical Electrical & Engineering, Nanjing University of Aeronautics and AstronauticsNanjing, China
Profilo Orcid
Cerca questo autore su
Sciendo|Google Scholar
Yu, Hu
, 
Xiaoyu Che

Xiaoyu Che

College of Mechanical Electrical & Engineering, Nanjing University of Aeronautics and AstronauticsNanjing, China
Cerca questo autore su
Sciendo|Google Scholar
Che, Xiaoyu
, 
Chao Zhang

Chao Zhang

College of Mechanical Electrical & Engineering, Nanjing University of Aeronautics and AstronauticsNanjing, China
Cerca questo autore su
Sciendo|Google Scholar
Zhang, Chao
, 
Rupeng Zhu

Rupeng Zhu

College of Mechanical Electrical & Engineering, Nanjing University of Aeronautics and AstronauticsNanjing, China
Cerca questo autore su
Sciendo|Google Scholar
Zhu, Rupeng
 e 
Weifang Chen

Weifang Chen

College of Mechanical Electrical & Engineering, Nanjing University of Aeronautics and AstronauticsNanjing, China
Cerca questo autore su
Sciendo|Google Scholar
Chen, Weifang
  
22 lug 2025
GGLCM: A Real-time Early Anomaly Detection Method for Mechanical Vibration Data with Missing Labels's Cover Image
INFORMAZIONI SU QUESTO ARTICOLO
Articolo precedente
Articolo Successivo
Cita
Scarica la copertina
Pubblicato online: 22 lug 2025
Pagine: 157 - 163
Ricevuto: 28 mag 2024
Accettato: 16 mag 2025
DOI: https://doi.org/10.2478/msr-2025-0019
Parole chiave
© 2025 Hu Yu et al., published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Fig. 1.

The diagram of the proposed GGLCM method.
The diagram of the proposed GGLCM method.

Fig. 2.

NGC overloaded gearbox fatigue life test bench.
NGC overloaded gearbox fatigue life test bench.

Fig. 3.

Schematic diagram of fault implantation.
Schematic diagram of fault implantation.

Fig. 4.

Fatigue life test procedure of a steel press gearbox.
Fatigue life test procedure of a steel press gearbox.

Fig. 5.

The temporal evolution of variable yTi y_T^i (the results of early anomaly detection on XJTU-SY). (a) Bearing1_1, (b) Bearing1_3, (c) Bearing2_2, (d) Bearing2_4, (e) Bearing3_1, (f) Bearing3_4.
The temporal evolution of variable yTi y_T^i (the results of early anomaly detection on XJTU-SY). (a) Bearing1_1, (b) Bearing1_3, (c) Bearing2_2, (d) Bearing2_4, (e) Bearing3_1, (f) Bearing3_4.

Fig. 6.

The detection result on the NGC dataset.
The detection result on the NGC dataset.

Fig. 7.

Performance comparison of three SOTA methods.
Performance comparison of three SOTA methods.

The detailed results for XJTU-SY datasets_

Subject Input ratio [%] YFactual [s] YFest [s] yFest*s {y^{F_{est}^*}}\left[ {\rm{s}} \right] Acc* [%]
+10 % +20 % +30 % +0 % +10 % +20 % +30 %
Bearing1_1 70 4 393.95 4 712.06 4 712.06 4 712.06 4 712.06 95.69 95.69 95.69 95.69
Bearing1_3 70 5 578.64 5 568.23 5 568.23 5 568.23 5 568.23 99.81 99.81 99.81 99.81
Bearing2_2 70 2 848.64 2 949.91 2 949.91 2 949.91 2 949.91 96.44 96.44 96.44 96.44
Bearing2_4 70 1 798.43 1 830.34 1 830.34 1 830.34 1 830.34 98.22 98.22 98.22 98.22
Bearing3_4 70 55 195.86 56 648.44 56 648.44 56 648.44 56 648.44 97.37 97.37 97.37 97.37
Bearing3_5 70 1 346.31 1 435.74 1 435.74 1 435.74 1 435.74 93.36 93.36 93.36 93.36

The parameters of the test gearbox_

Parameters Parameter values
Type MPGH2H900K63H
Rotation speed ratio 62.402
Rated power [kW] 710
Input rotation speed [rpm] 1 490
Maximum input torque [N·m] 10 032
Maximum output torque [kN·m] 626
Lubricating oil type ISO VG 320
Oil flow rate of lubrication station [l/min] 125
Oil mass [l] 230
Lingua:
Inglese
Frequenza di pubblicazione:
6 volte all'anno
Argomenti della rivista:
Ingegneria, Elettrotecnica, Ingegneria dell'automazione, metrologia e collaudo
Feed RSS della rivista