Scholarlink

In this paper, a deep learning model is used to forecast the geometric accuracy of the Single Point Incremental Forming (SPIF) process. To forecast the geometric accuracy of the components, a Stacked Auto-Encoder (SAE) network in conjunction with a backpropagation algorithm was selected. The springback phenomenon was chosen as a factor for measuring geometric precision. Nonetheless, six parameters were selected as input factors: wall angle, spindle speed, feed rate, vertical increment, tool trajectory, and sheet thickness. The novel proposed method is characterized by combining feature extraction and regression phases into a single model. Finally, the results show that the recommended strategy achieves better prediction accuracy for springback.

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

Application of Stacked Autoencoder Approach in SPIF Process for Springback Prediction

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sofien Akrichi,
  • Noureddine Ben Yahia

摘要

In this paper, a deep learning model is used to forecast the geometric accuracy of the Single Point Incremental Forming (SPIF) process. To forecast the geometric accuracy of the components, a Stacked Auto-Encoder (SAE) network in conjunction with a backpropagation algorithm was selected. The springback phenomenon was chosen as a factor for measuring geometric precision. Nonetheless, six parameters were selected as input factors: wall angle, spindle speed, feed rate, vertical increment, tool trajectory, and sheet thickness. The novel proposed method is characterized by combining feature extraction and regression phases into a single model. Finally, the results show that the recommended strategy achieves better prediction accuracy for springback.