Scholarlink

<p>Friction stir welding (FSW) for dissimilar materials of alloy and carbon fiber plays an important role in lightweight technology. Acoustic emission (AE) signals are generated in the FSW process, which contains information related to weld defects and weld joint forming quality. In this paper, an innovative feature extraction algorithm based on synchronous extraction transform optimized by Multi-resolution polynomial window function (MPSET) is proposed to optimize the unclear feature of AE signal for the defect monitoring of thermal assistance FSW (TA-FSW). TA-FSW experiments on Al alloy and carbon fiber reinforced thermoplastic (CFRTP) are carried out with AE signals measured. In MPSET, the window function is optimized by stochastic gradient descent (SGD) for optimization with maximum sidelobe attenuation as the objective. Then, a set of window functions is applied for synchronous feature extraction, followed by time–frequency rearrangement. The MPSET is analyzed for simulation signals and AE signals compared with five other popular time–frequency algorithms. The experimental results indicate that spectrum aliasing is avoided in the MPSET spectrum, and the EECR is increased by 20% compared to other algorithms. The multi-resolution window function limits spectral leakage and improves spectral resolution in the time–frequency transformation, which exhibits the primary frequency band highlighted. Moreover, the relation between the MPSET spectrum and welding defects is explored. The primary frequency of the AE signal is 18&#xa0;kHz in TA-FSW for Al-CFRTP. The frequency band transfer to 15&#xa0;kHz is accompanied by hole and tunnel defects, and the power transfer is related to the size of the defect. Also, the frequency band transfer to 23&#xa0;kHz depends on defect distances from the Al alloy body. The MPSET provides an effective time–frequency algorithm for feature extraction of TA-FSW defects.</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

The feature extraction algorithm of AE signal based on SET optimized by MPWF for TA-FSW defect monitoring

  • 期刊论文

  • 发表日期：

    2026年2月26日

查看全文
  • Haiwei Long,
  • Yibo Sun,
  • Yanming Xing,
  • Libin Fu,
  • Siwen Liang

摘要

Friction stir welding (FSW) for dissimilar materials of alloy and carbon fiber plays an important role in lightweight technology. Acoustic emission (AE) signals are generated in the FSW process, which contains information related to weld defects and weld joint forming quality. In this paper, an innovative feature extraction algorithm based on synchronous extraction transform optimized by Multi-resolution polynomial window function (MPSET) is proposed to optimize the unclear feature of AE signal for the defect monitoring of thermal assistance FSW (TA-FSW). TA-FSW experiments on Al alloy and carbon fiber reinforced thermoplastic (CFRTP) are carried out with AE signals measured. In MPSET, the window function is optimized by stochastic gradient descent (SGD) for optimization with maximum sidelobe attenuation as the objective. Then, a set of window functions is applied for synchronous feature extraction, followed by time–frequency rearrangement. The MPSET is analyzed for simulation signals and AE signals compared with five other popular time–frequency algorithms. The experimental results indicate that spectrum aliasing is avoided in the MPSET spectrum, and the EECR is increased by 20% compared to other algorithms. The multi-resolution window function limits spectral leakage and improves spectral resolution in the time–frequency transformation, which exhibits the primary frequency band highlighted. Moreover, the relation between the MPSET spectrum and welding defects is explored. The primary frequency of the AE signal is 18 kHz in TA-FSW for Al-CFRTP. The frequency band transfer to 15 kHz is accompanied by hole and tunnel defects, and the power transfer is related to the size of the defect. Also, the frequency band transfer to 23 kHz depends on defect distances from the Al alloy body. The MPSET provides an effective time–frequency algorithm for feature extraction of TA-FSW defects.