EUVIMEDEUVIMED entdecken
Crossref

A fault diagnosis method for the fluid end of drilling pump with adaptive enhancement of weak fault features

Gang Li, Dalong Hu, Xiucheng Guo, Jiaxing Ao, Jiaming Shen, Ruijun Liu

Measurement Science and Technology · 30.04.2025 · 36 (4) · 046101

DOI: 10.1088/1361-6501/adbb0f ↗

Kurz gefasst

Fachartikel in der Zeitschrift Measurement Science and Technology, online erschienen am 14.03.2025. Verfasst von Gang Li und 5 weiteren Autor:innen. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 42 Literaturangaben. Gefördert von: Open Fund of Technology Innovation Center for Directional Drilling Engineering of Ministry of Natural Resources, Nanchong City & Southwest Petroleum University Science and Technology Strategic Cooperation Special Fund.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The drilling pump is the core circulation equipment used in oil and gas drilling. Timely diagnosis and troubleshooting of the suction and discharge valves are essential for preventing fault escalation and ensuring the smooth operation of the drilling pump. However, in real working conditions, these valves in the fluid end can experience various types of faults, often obscured by significant noise interference, and current diagnostic methods lack effectiveness. This study focuses on a five-cylinder drilling pump, investigating the fault states of the suction and discharge valves under actual working conditions. A data-driven fault diagnosis method for the fluid end is proposed, introducing an innovative adaptive maximum cyclostationarity blind deconvolution (PCYCBD) signal processing algorithm based on cumulative power spectral density. Additionally, a multi-scale convolutional neural network model, MSShuffleNet, is developed to enhance diagnostic accuracy across different batches, with fault diagnosis experiments conducted on the fluid end. The results show that the PCYCBD algorithm effectively reduces noise interference, improving signal quality, while the MSShuffleNet model efficiently extracts fault features at multiple scales. The proposed method achieves an average accuracy of 98.96% across eight fault states. Furthermore, the Mechanical Failure Prevention Technology Association (MFPT) bearing fault dataset was used to evaluate the method’s adaptability, achieving a diagnostic accuracy of 99.78%, demonstrating its robustness and generalization capabilities. The method provides a fast and accurate solution for diagnosing fluid end faults in drilling pumps under noisy conditions and offers a valuable reference for diagnosing faults in other critical components.

Förderung

  • Open Fund of Technology Innovation Center for Directional Drilling Engineering of Ministry of Natural Resources
  • Nanchong City & Southwest Petroleum University Science and Technology Strategic Cooperation Special Fund

Zitieren

APA

Li, G., Hu, D., Guo, X., Ao, J., Shen, J., & Liu, R. (2025). A fault diagnosis method for the fluid end of drilling pump with adaptive enhancement of weak fault features. Measurement Science and Technology, 36(4), 046101. https://doi.org/10.1088/1361-6501/adbb0f

Vancouver

Li G, Hu D, Guo X, Ao J, Shen J, Liu R. A fault diagnosis method for the fluid end of drilling pump with adaptive enhancement of weak fault features. Measurement Science and Technology. 2025;36(4):046101. doi:10.1088/1361-6501/adbb0f

BibTeX
@article{Li2025,
  author = {Li, Gang and Hu, Dalong and Guo, Xiucheng and Ao, Jiaxing and Shen, Jiaming and Liu, Ruijun},
  title = {A fault diagnosis method for the fluid end of drilling pump with adaptive enhancement of weak fault features},
  journal = {Measurement Science and Technology},
  year = {2025},
  volume = {36},
  number = {4},
  pages = {046101},
  doi = {10.1088/1361-6501/adbb0f}
}
Kostenlos starten →