Scholarlink

<p>Cutting tools are crucial in manufacturing industries for precise machining. Cutting tools’ chatter is an undesirable self-excited vibration that negatively affecting surface quality and productivity. Moreover, tool wear occurs progressively due to friction between the tool and the workpiece. Therefore, monitoring chatter and tool wear is pivotal for maintaining machining processes’ quality, efficiency, safety, and cost-effectiveness. Hence, this research aims to enhance the monitoring accuracy of chatter and tool wear state (TWS) classification by proposing some key methodologies after data collection, such as normalization, optimized variational mode decomposition (VMD) by genetic algorithm, intrinsic mode functions (IMFs) selection, spectrogram image extraction and CNN-MobileNet model to show the comprehensive and innovative approach for cutting tools. This research utilized four stick-out conditions for chatter and three cutters for tool wear for training and validation purposes, where chatter and TWS of cutting tools have four classes. It is found that there is a need to optimize the parameters of VMD and select the sensitive IMFs to achieve the best performance. This paper uses spectral and Renyi entropy as fitness functions to optimize VMD parameters. Additionally, it introduces a novel ECK Venn-based IMFs selection strategy, combining energy, correlation, and Kullback-Leibler distance. Hence, this process enhances signal representation during reconstruction. Lastly, spectrogram images provide intuitive visualizations and employ the CNN-MobileNet model to ensure accurate chatter and tool wear state classification. The novelty of this study lies in its utilization of an optimized VMD-ECK-Venn-based approach for selecting dominant IMFs, followed by a spectrogram to generate images for input into a MobileNet deep learning architecture. The findings of this research demonstrate the effectiveness of the proposed methodology for chatter and TWS classification in cutting tools. The proposed methodology shows its superiority over traditional methods and achieves fault classification accuracy in between 97 %-100 % and minimum validation loss is 4.2944×10<sup>−8</sup>.</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

Optimized VMD and novel ECK-Venn based approach to IMF selection for chatter and tool wear classification

  • 期刊论文

  • 发表日期：

    2025年1月6日

查看全文
  • Neha Patel,
  • Rajiv Nandan Rai

摘要

Cutting tools are crucial in manufacturing industries for precise machining. Cutting tools’ chatter is an undesirable self-excited vibration that negatively affecting surface quality and productivity. Moreover, tool wear occurs progressively due to friction between the tool and the workpiece. Therefore, monitoring chatter and tool wear is pivotal for maintaining machining processes’ quality, efficiency, safety, and cost-effectiveness. Hence, this research aims to enhance the monitoring accuracy of chatter and tool wear state (TWS) classification by proposing some key methodologies after data collection, such as normalization, optimized variational mode decomposition (VMD) by genetic algorithm, intrinsic mode functions (IMFs) selection, spectrogram image extraction and CNN-MobileNet model to show the comprehensive and innovative approach for cutting tools. This research utilized four stick-out conditions for chatter and three cutters for tool wear for training and validation purposes, where chatter and TWS of cutting tools have four classes. It is found that there is a need to optimize the parameters of VMD and select the sensitive IMFs to achieve the best performance. This paper uses spectral and Renyi entropy as fitness functions to optimize VMD parameters. Additionally, it introduces a novel ECK Venn-based IMFs selection strategy, combining energy, correlation, and Kullback-Leibler distance. Hence, this process enhances signal representation during reconstruction. Lastly, spectrogram images provide intuitive visualizations and employ the CNN-MobileNet model to ensure accurate chatter and tool wear state classification. The novelty of this study lies in its utilization of an optimized VMD-ECK-Venn-based approach for selecting dominant IMFs, followed by a spectrogram to generate images for input into a MobileNet deep learning architecture. The findings of this research demonstrate the effectiveness of the proposed methodology for chatter and TWS classification in cutting tools. The proposed methodology shows its superiority over traditional methods and achieves fault classification accuracy in between 97 %-100 % and minimum validation loss is 4.2944×10−8.