Scholarlink

<p>Continuous monitoring of the operational status of gear transmission systems is increasingly crucial. To achieve this, the use of suitable monitoring techniques is essential, among which vibration analysis stands out. This paper combines vibration data with an artificial neural network (ANN) to automate the identification of gear defects. In our research data, it is proved that the ANN has high precision of classifying the defects of gears (tooth breakage, half-tooth breakage, and tooth wear) by the indication of a low mean square error (MSE) of 5.2228e<sup>−22</sup>, after 2000 iterations. The time-based counters (RMS, VAR, STD) showed a linear relationship to rotational speed and defect severity, and the frequency domain analysis indicated that some of the highest fault sensitivities belonged to the axial and horizontal vibrations. The above outcomes confirm the trustworthiness of hopes in ANN towards preventative maintenance, and serves a powerful tool of early detection of faults within gear systems. The promising results of this approach highlight its potential for accurate defect identification, thus enhancing the reliability of preventive maintenance operations in these critical systems.</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

Classification of defects in a gear system by applying artificial neural networks RNA

  • 期刊论文

  • 发表日期：

    2025年8月14日

查看全文
  • Lias Rehai,
  • Ahmed Bellaouar,
  • Rachid Chaib,
  • Nabil Talbi,
  • Farhan Lafta Rashid,
  • Mohamed Kezzar,
  • Ibrahim Mahariq

摘要

Continuous monitoring of the operational status of gear transmission systems is increasingly crucial. To achieve this, the use of suitable monitoring techniques is essential, among which vibration analysis stands out. This paper combines vibration data with an artificial neural network (ANN) to automate the identification of gear defects. In our research data, it is proved that the ANN has high precision of classifying the defects of gears (tooth breakage, half-tooth breakage, and tooth wear) by the indication of a low mean square error (MSE) of 5.2228e−22, after 2000 iterations. The time-based counters (RMS, VAR, STD) showed a linear relationship to rotational speed and defect severity, and the frequency domain analysis indicated that some of the highest fault sensitivities belonged to the axial and horizontal vibrations. The above outcomes confirm the trustworthiness of hopes in ANN towards preventative maintenance, and serves a powerful tool of early detection of faults within gear systems. The promising results of this approach highlight its potential for accurate defect identification, thus enhancing the reliability of preventive maintenance operations in these critical systems.