Scholarlink

Background <p>During the monitoring of rolling bearing status, there are problems such as complex working conditions and significant environmental interference in data collection, causing abnormal data to appear. Correcting abnormal data is of great significance for ensuring the accurate extraction information of rolling bearing status because of the high repetitive costs in actual operation.</p> Purpose <p>It is necessary to retain as much valid information as possible from the data with anomalies. This paper studies the problem of removing abnormal data. A correction method based on Super smoother and compressive sensing (CS) is proposed to eliminate random abnormal impulse interference in vibration signals.</p> Methods <p>Firstly, a vibration data fitting method based on Super smoother is introduced, and the amplitude distribution characteristics of the vibration signal are studied. Then a random abnormal data cleaning and repair method based on the CS principle is proposed. Afterwards the Super smoother and compressive sensing (SCS) are put forward to correct abnormal data, along with the construction of a comprehensive evaluation index for correcting capability. Lastly, multiple experiments are conducted to validate the effectiveness of the proposed method.</p> Results <p>Through cleaning data from multiple experiments, the results reveal that the SCS algorithm can smooth vibration signals, fit signal trends with error sets, eliminate abnormal data using 3σ criteria, and effectively correct outliers. It can effectively correct single or multiple abnormal impact signals. The comprehensive evaluation index of the proposed method is the highest, and it is effective to analyze and verify the data-correcting ability of various methods.</p> Conclusion <p>The time–frequency domain features of the data corrected by the SCS method are closer to normal signals. For both constant speed and variable speed bearing signals, the proposed method can effectively correct abnormal signals. Stress test indicates that the method has stability and superiority. The proposed method can improve data quality and facilitate subsequent utilization, reduces the difficulty of extracting fault features of rolling bearings, and is of great significance for fault diagnosis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Abnormal Vibration Data Correction of Rolling Bearings Based on Super Smoother and Compressive Sensing

  • 期刊论文

  • 发表日期：

    2025年8月5日

查看全文
  • Haiming Wang,
  • Yongqiang Liu,
  • Shaopu Yang,
  • Wenpeng Liu,
  • Yanli Wang

摘要

Background

During the monitoring of rolling bearing status, there are problems such as complex working conditions and significant environmental interference in data collection, causing abnormal data to appear. Correcting abnormal data is of great significance for ensuring the accurate extraction information of rolling bearing status because of the high repetitive costs in actual operation.

Purpose

It is necessary to retain as much valid information as possible from the data with anomalies. This paper studies the problem of removing abnormal data. A correction method based on Super smoother and compressive sensing (CS) is proposed to eliminate random abnormal impulse interference in vibration signals.

Methods

Firstly, a vibration data fitting method based on Super smoother is introduced, and the amplitude distribution characteristics of the vibration signal are studied. Then a random abnormal data cleaning and repair method based on the CS principle is proposed. Afterwards the Super smoother and compressive sensing (SCS) are put forward to correct abnormal data, along with the construction of a comprehensive evaluation index for correcting capability. Lastly, multiple experiments are conducted to validate the effectiveness of the proposed method.

Results

Through cleaning data from multiple experiments, the results reveal that the SCS algorithm can smooth vibration signals, fit signal trends with error sets, eliminate abnormal data using 3σ criteria, and effectively correct outliers. It can effectively correct single or multiple abnormal impact signals. The comprehensive evaluation index of the proposed method is the highest, and it is effective to analyze and verify the data-correcting ability of various methods.

Conclusion

The time–frequency domain features of the data corrected by the SCS method are closer to normal signals. For both constant speed and variable speed bearing signals, the proposed method can effectively correct abnormal signals. Stress test indicates that the method has stability and superiority. The proposed method can improve data quality and facilitate subsequent utilization, reduces the difficulty of extracting fault features of rolling bearings, and is of great significance for fault diagnosis.