Welcome to our website!
Consultation hotline: RSS EMAIL-ALERT
Research on Improved Fault Diagnosis Method of Hydraulic Pump Based on Optimal NAS
CSTR:
Author:
Affiliation:

Clc Number:

Fund Project:

  • Article
  • |
  • Figures
  • |
  • Metrics
  • |
  • Reference
  • |
  • Related
  • |
  • Cited by
  • |
  • Materials
  • |
  • Comments
    Abstract:

    In view of the low performance evaluation efficiency of neural architecture search (NAS) method in searching optimal structure,and the low diagnostic accuracy of hydraulic pump fault diagnosis due to poor model generalization ability,an improved Data-free NAS method was proposed.CAME optimizer and cosine annealing with warm restart algorithm were introduced to replace SGD optimizer and LambdaLR optimization algorithm,respectively,to improve the performance estimation and verification functions of Data-free NAS such as diagnostic accuracy and computational efficiency.Through the verified analysis of the hydraulic pump fault measured fault experiment,it can be seen that the improved method has obvious effectiveness and superiority over the original method;the CAME optimizer has obvious advantages in optimizing hyperparameters such as learning rate and momentum weight of the network,with accuracy and efficiency improved by 7.24% and 37.5%,respectively,with an accuracy up to 100%; the learning rate parameters can be improved by cosine annealing with warm restart algorithm,and the efficiency is improved by 81.25%.

    Reference
    Related
    Cited by
Get Citation

郑直,刘彤谣,赵文博,刘伟民,王志军.基于最优架构搜索网络的液压泵故障诊断改进方法研究[J].机床与液压英文版,2024,52(19):216-224.

Copy
Share
Article Metrics
  • Abstract:
  • PDF:
  • HTML:
  • Cited by:
History
  • Received:
  • Revised:
  • Adopted:
  • Online: October 23,2024
  • Published: October 15,2024
Article QR Code