TY - GEN
T1 - Composite Fault Diagnosis of Rotating Machinery With Collaborative Learning
AU - Pavan Kumar, M. P.
AU - Tang, Cheng Jyun
AU - Chen, Kun Chih Jimmy
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Rotating machinery is a machine with a rotating component to do the energy transformation, which is widely used in vehicle engines, power plants, manufacturing factories, etc. Because of the continuous rotation, the bearing and gear are subject to a defect in the rotating machines, which damages the reliability of the machine. To provide a sustained normal working status, predictive maintenance (PdM) is usually applied to monitor the working health during the machine running by gathering different sensing data. However, highly diverse and massive sensing data increase the challenge to analyze the fault signals in the machine. Besides, the situation with composite faults (i.e., faults that happen in different components in a machine) worsens the difficulty to diagnose the target machine efficiently. To solve this problem, we propose a kind of hierarchy collaborative learning method in this work. Different from conventional centralized learning to analyze heterogeneous sensing data, collaborative learning uses multiple sub-learning units to analyze the local homogeneous sensing data priorly. Then, perform fusion operation for each sub learning unit obtained from homogenous sensing data to predict the final composite fault diagnosis results. In this way, we can not only ensure the quality of the signal fault diagnosis but improve the accuracy of the composite fault's diagnosis significantly. Compared with the centralized learning methods, the proposed collaborative learning method can achieve a 12% improvement in the accuracy of predicting the composite fault diagnosis.
AB - Rotating machinery is a machine with a rotating component to do the energy transformation, which is widely used in vehicle engines, power plants, manufacturing factories, etc. Because of the continuous rotation, the bearing and gear are subject to a defect in the rotating machines, which damages the reliability of the machine. To provide a sustained normal working status, predictive maintenance (PdM) is usually applied to monitor the working health during the machine running by gathering different sensing data. However, highly diverse and massive sensing data increase the challenge to analyze the fault signals in the machine. Besides, the situation with composite faults (i.e., faults that happen in different components in a machine) worsens the difficulty to diagnose the target machine efficiently. To solve this problem, we propose a kind of hierarchy collaborative learning method in this work. Different from conventional centralized learning to analyze heterogeneous sensing data, collaborative learning uses multiple sub-learning units to analyze the local homogeneous sensing data priorly. Then, perform fusion operation for each sub learning unit obtained from homogenous sensing data to predict the final composite fault diagnosis results. In this way, we can not only ensure the quality of the signal fault diagnosis but improve the accuracy of the composite fault's diagnosis significantly. Compared with the centralized learning methods, the proposed collaborative learning method can achieve a 12% improvement in the accuracy of predicting the composite fault diagnosis.
UR - http://www.scopus.com/inward/record.url?scp=85130503896&partnerID=8YFLogxK
U2 - 10.1109/VLSI-DAT54769.2022.9768050
DO - 10.1109/VLSI-DAT54769.2022.9768050
M3 - Conference contribution
AN - SCOPUS:85130503896
T3 - 2022 International Symposium on VLSI Design, Automation and Test, VLSI-DAT 2022 - Proceedings
BT - 2022 International Symposium on VLSI Design, Automation and Test, VLSI-DAT 2022 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2022 International Symposium on VLSI Design, Automation and Test, VLSI-DAT 2022
Y2 - 18 April 2022 through 21 April 2022
ER -