Mechanical fault diagnosis of high voltage circuit breaker using multimodal data fusion

View article
PeerJ Computer Science

Main article text

 

Introduction

  • (1) A high voltage circuit breaker fault detection platform was constructed, where vibration signals collected from four different locations on the breaker effectively identify contact faults, mechanism seizures, bolt loosening, and spring fatigue.

  • (2) A simple and effective CANet was developed, utilizing the combination of one-dimensional convolution and self-attention mechanisms to efficiently extract key features from multi-point time series data and classify faults.

  • (3) Multi-point data fusion was implemented, combining vibration signals from various locations on the high voltage circuit breaker, overcoming the limitations of traditional fault diagnosis methods that rely on a single signal source, and enhancing the accuracy and reliability of fault diagnosis.

Materials and Methods

Signal acquisition

Convolutional attention network

CA module

where h denotes the number of heads, headi represents the output of the ith head, and Wo is the output transformation matrix. WQi, WKi, WVi are the transformation matrices for query, key, and value, respectively, and dk is the dimensionality of the key tensor.

Results

Implementation details

Evaluation metrics

Experiments

Discussion

Conclusions

  • (1) We constructed a platform for collecting mechanical fault data of HVCBs using vibration sensors. By using fault injection methods to expand the dataset, we have overcome the challenges associated with the scarcity of fault data.

  • (2) We developed a simple yet effective CANet, utilizing one-dimensional convolution and an attention mechanism for feature extraction from multi-point data. The capability of CANet to process and analyze multi-point data has been proven to be highly effective.

  • (3) We have validated the capability of the attention mechanism in processing multi-point time-series data. Our findings demonstrate that the attention mechanism is a key factor in enhancing the diagnostic capabilities of CANet, not only improving overall accuracy but also increasing the precision of the model in identifying various types of faults.

Supplemental Information

Training model code.

DOI: 10.7717/peerj-cs.2248/supp-1

Main function code.

DOI: 10.7717/peerj-cs.2248/supp-2

Transformer code.

DOI: 10.7717/peerj-cs.2248/supp-3

The code for preprocessing the dataset.

DOI: 10.7717/peerj-cs.2248/supp-4

Model building code.

DOI: 10.7717/peerj-cs.2248/supp-5

Additional Information and Declarations

Competing Interests

Tianhui Li, Yanwei Xia, Xianhai Pang, Chaomin Gu, Chi Dong and Shijie Lu are employed by State Grid Hebei Electric Power Research Institute and State Grid Hebei Energy Technology Service Co., Ltd. Jihong Zhu is employed by Design and Development Department, Nanjing Hz Electric Co., Ltd. Hui Fan and Li Zhen are employed by State Grid Hebei Electric Power Supply Co., Ltd. The authors declare that they have no competing interests.

Author Contributions

Tianhui Li conceived and designed the experiments, authored or reviewed drafts of the article, and approved the final draft.

Yanwei Xia conceived and designed the experiments, authored or reviewed drafts of the article, and approved the final draft.

Xianhai Pang conceived and designed the experiments, analyzed the data, authored or reviewed drafts of the article, and approved the final draft.

Jihong Zhu conceived and designed the experiments, analyzed the data, authored or reviewed drafts of the article, and approved the final draft.

Hui Fan conceived and designed the experiments, analyzed the data, performed the computation work, prepared figures and/or tables, and approved the final draft.

Li Zhen analyzed the data, performed the computation work, prepared figures and/or tables, and approved the final draft.

Chaomin Gu performed the experiments, analyzed the data, performed the computation work, prepared figures and/or tables, and approved the final draft.

Chi Dong performed the experiments, performed the computation work, prepared figures and/or tables, and approved the final draft.

Shijie Lu performed the experiments, prepared figures and/or tables, and approved the final draft.

Data Availability

The following information was supplied regarding data availability:

The code is available in the Supplemental Files.

The data and code are available at GitHub and Zenodo:

- https://github.com/tianhuiLi700/CAnet.

- https://doi.org/10.5281/zenodo.13291531.

Funding

This work was supported by the State Grid Hebei Electric Power Co., Ltd. Technology Project Funding (kj2022-062) and the National Natural Science Foundation of China (No. 62371253 and No. 52278119). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

1 Citation 705 Views 34 Downloads

Your institution may have Open Access funds available for qualifying authors. See if you qualify

Publish for free

Comment on Articles or Preprints and we'll waive your author fee
Learn more

Five new journals in Chemistry

Free to publish • Peer-reviewed • From PeerJ
Find out more