Scholarlink

Abstract <p>With the advancements in machine learning and theemergence of intelligent manufacturing, there is an increasing needto develop reliable and robust monitoring techniques forultraprecision machining (UPM) status. UPM is crucial formanufacturing high-precision components such as miniature lenses,aspheric mirrors, freeform surfaced optics, and structural surfaces.Traditional monitoring methods in UPM involve using multiplesensors, which can be costly and time-consuming in operations.Consequently, there is a need for optimal feature extraction andclassification techniques in-process, which are applicable inparticular. Feature extraction is vital in extracting valuableinformation from data without losing relevant details. The currentstudy uses single-axis vibrational signal-based Mel-frequencycepstral coefficient (MFCC) features to tackle this challenge. TheseMFCC features are used as the input for the Machine Learning (ML)based classification algorithms. The findings of this experimentalstudy demonstrate that the combination of MFCC features and MLalgorithms can reliably predict and monitor machining status, withan average accuracy of 94.64% by the machine learning models at 30filters. Additionally, the proposed MFCC model integrated with MLexhibits excellent performance and will be useful for the monitoringof the process anomalies during Diamond turning. This ability topredict and monitor the machining status is expected to holdsignificant value, particularly as the fabrication of freeformsurfaced optics and devices, in which the ultraprecisionmanufacturing processes become increasingly complex andtime-consuming.</p> Graphical abstract <p></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

In-process monitoring and prediction of machining status in ultraprecision diamond turning using Mel-frequency cepstral coefficient approach combined with machine learning

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • K. Manjunath,
  • Suman Tewary,
  • Neha Khatri,
  • Kai Cheng

摘要

Abstract

With the advancements in machine learning and theemergence of intelligent manufacturing, there is an increasing needto develop reliable and robust monitoring techniques forultraprecision machining (UPM) status. UPM is crucial formanufacturing high-precision components such as miniature lenses,aspheric mirrors, freeform surfaced optics, and structural surfaces.Traditional monitoring methods in UPM involve using multiplesensors, which can be costly and time-consuming in operations.Consequently, there is a need for optimal feature extraction andclassification techniques in-process, which are applicable inparticular. Feature extraction is vital in extracting valuableinformation from data without losing relevant details. The currentstudy uses single-axis vibrational signal-based Mel-frequencycepstral coefficient (MFCC) features to tackle this challenge. TheseMFCC features are used as the input for the Machine Learning (ML)based classification algorithms. The findings of this experimentalstudy demonstrate that the combination of MFCC features and MLalgorithms can reliably predict and monitor machining status, withan average accuracy of 94.64% by the machine learning models at 30filters. Additionally, the proposed MFCC model integrated with MLexhibits excellent performance and will be useful for the monitoringof the process anomalies during Diamond turning. This ability topredict and monitor the machining status is expected to holdsignificant value, particularly as the fabrication of freeformsurfaced optics and devices, in which the ultraprecisionmanufacturing processes become increasingly complex andtime-consuming.

Graphical abstract