Scholarlink

Wire arc additive manufacturing (WAAM)Wire Arc Additive Manufacturing (WAAM) has strong potential to repairRepair damaged or fractured metal components. The damage creates random irregular surfaces, which makes the repairRepair difficult with conventional welding-based deposition. This paper presents an adaptive, monitoring-based method for WAAMWire Arc Additive Manufacturing (WAAM) repairRepair using non-planar slicing. A one-dimensional laser displacement sensorLaser displacement sensor performs a raster scan to capture the damaged surface profile. A long short-term memory (LSTM) model uses that profile and recent deposition history to predict local travel speeds needed to obtain the target bead geometry. Predictions are sent to an in-house hardware interfaceInterface that regulates real-time robot travel speed. The combined method aims to reduce pre-machining, preserve sound substrate, and minimise stair-stepping on curved and inclined surfaces. Experimental trials compare the proposed methodology with conventional planar repairRepair. The controlled deposition significantly decreased the standard deviation of the top layer from 1.65 to 1.02 mm compared to deposition without control. The demonstration for repairRepair also resulted in a minimal RMSE of 0.38 mm while effectively compensating for the uneven damaged surface. Results also show that ML-assisted, non-planar adaptive slicing smooths irregular surfaces over a few layers and improves repairRepair efficiency. The method offers a practical route to industrial WAAMWire Arc Additive Manufacturing (WAAM) repairRepair and reduces operational downtime significantly.

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

Machine Learning Based Adaptive Deposition Control for Wire Arc Additive Manufacturing Repair

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Anas Ullah Khan,
  • V. Ajay,
  • Amber Shrivastava

摘要

Wire arc additive manufacturing (WAAM)Wire Arc Additive Manufacturing (WAAM) has strong potential to repairRepair damaged or fractured metal components. The damage creates random irregular surfaces, which makes the repairRepair difficult with conventional welding-based deposition. This paper presents an adaptive, monitoring-based method for WAAMWire Arc Additive Manufacturing (WAAM) repairRepair using non-planar slicing. A one-dimensional laser displacement sensorLaser displacement sensor performs a raster scan to capture the damaged surface profile. A long short-term memory (LSTM) model uses that profile and recent deposition history to predict local travel speeds needed to obtain the target bead geometry. Predictions are sent to an in-house hardware interfaceInterface that regulates real-time robot travel speed. The combined method aims to reduce pre-machining, preserve sound substrate, and minimise stair-stepping on curved and inclined surfaces. Experimental trials compare the proposed methodology with conventional planar repairRepair. The controlled deposition significantly decreased the standard deviation of the top layer from 1.65 to 1.02 mm compared to deposition without control. The demonstration for repairRepair also resulted in a minimal RMSE of 0.38 mm while effectively compensating for the uneven damaged surface. Results also show that ML-assisted, non-planar adaptive slicing smooths irregular surfaces over a few layers and improves repairRepair efficiency. The method offers a practical route to industrial WAAMWire Arc Additive Manufacturing (WAAM) repairRepair and reduces operational downtime significantly.