Scholarlink

<p>Wire arc additive manufacturing (WAAM) is a directed energy deposition technology that offers great potential for real-time manufacturing of ocean-going vessels. However, the complex maritime environment poses challenges to forming quality due to ship-induced motion. This study presents a method for predicting the forming quality of WAAM under variable shipboard conditions by analyzing current signals. Using time–frequency analysis and machine learning, a predictive model was developed to address forming deviations. Wavelet transform was applied to extract time–frequency features from current signals, and the retentive networks meet vision transformers (RMT) model was implemented for quality prediction. Experimental results reveal that the RMT model outperforms the traditional support vector regression (SVR) model in both stable and variable sea conditions, achieving lower prediction errors (RMSE of 0.0388, MAE of 0.0226), compared to SVR's RMSE of 0.0547 and MAE of 0.0428. Additionally, t-SNE analysis visualized the predictions, confirming the RMT model's robustness and effectiveness in dynamic environments. These findings provide both theoretical and practical insights for advancing shipboard WAAM applications under challenging real-world conditions.</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

Prediction of Forming Quality in Shipboard Wire Arc Additive Manufacturing Based on Current Signal Analysis

  • 期刊论文

  • 发表日期：

    2025年7月12日

查看全文
  • Zhun Wang,
  • Zilong Wang,
  • Petro Pavlenko,
  • Zhenhua Li,
  • Jinbao Wang,
  • Xuezhi Shi

摘要

Wire arc additive manufacturing (WAAM) is a directed energy deposition technology that offers great potential for real-time manufacturing of ocean-going vessels. However, the complex maritime environment poses challenges to forming quality due to ship-induced motion. This study presents a method for predicting the forming quality of WAAM under variable shipboard conditions by analyzing current signals. Using time–frequency analysis and machine learning, a predictive model was developed to address forming deviations. Wavelet transform was applied to extract time–frequency features from current signals, and the retentive networks meet vision transformers (RMT) model was implemented for quality prediction. Experimental results reveal that the RMT model outperforms the traditional support vector regression (SVR) model in both stable and variable sea conditions, achieving lower prediction errors (RMSE of 0.0388, MAE of 0.0226), compared to SVR's RMSE of 0.0547 and MAE of 0.0428. Additionally, t-SNE analysis visualized the predictions, confirming the RMT model's robustness and effectiveness in dynamic environments. These findings provide both theoretical and practical insights for advancing shipboard WAAM applications under challenging real-world conditions.