Scholarlink

Background <p>Todays, using of rotating machineries are very popular in industry centers. Since these machineries are used as moving or dynamic parts of various systems, they encounter different types of faults. If these faults are not fixed properly, they can cause failures with costly problems. One of the major defects in rotating machineries is unbalance.</p> Methods <p>In this research, it is attempted to classification and predict unbalance parameters by Empirical Mode Decomposition (EMD) and Random Forest (RF) algorithm with higher accuracy and simple way. A rotating test rig with four disks capable of placing different weights at different distances from the shaft center has been used in this study. The vibrational results were recorded for each of each test separately. The accelerations of the bearings, obtained from these results, were processed using empirical mode decomposition and then used as inputs for the random forest algorithm.</p> Results <p>After training the algorithms comparing the predicted values with the empirical data indicates the effectiveness of these method in determining the location and severity of the fault.</p> Conclusion <p>The accuracy of hybridizing EMD and RF show acceptable accuracy in locating of unbalance position in terms of disk number (75% accuracy), mass (82%) and radius of eccentricity (88%).</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 and Diagnosis of the Rotor Unbalance Parameters via Hybridized EMD and RF

  • 期刊论文

  • 发表日期：

    2025年3月25日

查看全文
  • Mohammad Gohari,
  • Mohammad Hossein Ghorbani

摘要

Background

Todays, using of rotating machineries are very popular in industry centers. Since these machineries are used as moving or dynamic parts of various systems, they encounter different types of faults. If these faults are not fixed properly, they can cause failures with costly problems. One of the major defects in rotating machineries is unbalance.

Methods

In this research, it is attempted to classification and predict unbalance parameters by Empirical Mode Decomposition (EMD) and Random Forest (RF) algorithm with higher accuracy and simple way. A rotating test rig with four disks capable of placing different weights at different distances from the shaft center has been used in this study. The vibrational results were recorded for each of each test separately. The accelerations of the bearings, obtained from these results, were processed using empirical mode decomposition and then used as inputs for the random forest algorithm.

Results

After training the algorithms comparing the predicted values with the empirical data indicates the effectiveness of these method in determining the location and severity of the fault.

Conclusion

The accuracy of hybridizing EMD and RF show acceptable accuracy in locating of unbalance position in terms of disk number (75% accuracy), mass (82%) and radius of eccentricity (88%).