Predictive Model for Deposition Success in Wire Laser Additive Manufacturing

Abstract

Wire alignment is a consistent challenge in wire laser additive manufacturing (WLAM). The position of the wire before the start of deposition could decide the quality and deposition success. Convolutional neural network (CNN) is known for providing excellent results in image classification. The image dataset consisted of 1422 images, which were split into a ratio of 60:20:20, representing the training, validation, and testing data. The images were broadly categorised as Success or Failure, referring to the estimation of the possibility of a smooth and defect-free deposition. The proposed model was evaluated on various metrics and yielded an accuracy of 90.06%. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.