Abstract:Directed Energy Deposition (DED) offers significant potential for manufacturing complex and multi-material parts. However, internal defects such as porosity and cracks can compromise mechanical properties and overall performance. This study focuses on in-situ monitoring and characterization of melt pools associated with porosity, aiming to improve defect detection and quality control in DED-printed parts. Traditional machine learning approaches for defect identification rely on extensive labeled datasets, often scarce and expensive to generate in real-world manufacturing. To address this, our framework employs self-supervised learning on unlabeled melt pool data using a Vision Transformer-based Masked Autoencoder (MAE) to produce highly representative embeddings. These fine-tuned embeddings are leveraged via transfer learning to train classifiers on a limited labeled dataset, enabling the effective identification of melt pool anomalies. We evaluate two classifiers: (1) a Vision Transformer (ViT) classifier utilizing the fine-tuned MAE Encoder's parameters and (2) the fine-tuned MAE Encoder combined with an MLP classifier head. Our framework achieves overall accuracy ranging from 95.44% to 99.17% and an average F1 score exceeding 80%, with the ViT Classifier slightly outperforming the MAE Encoder Classifier. This demonstrates the scalability and cost-effectiveness of our approach for automated quality control in DED, effectively detecting defects with minimal labeled data.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: in the Directed Energy Deposition (DED) manufacturing process, how to perform in - situ molten pool characterization by thermal imaging technology to detect internal defects (such as pores). Specifically, the research aims to improve the accuracy of defect detection and quality control, especially in the case of a lack of a large amount of labeled data. ### Problem Background DED is an advanced additive manufacturing technology, which is widely used in the aerospace and biomedical fields. However, internal defects such as lack of fusion, pores and cracks are easily generated during the DED process, and these defects will damage the mechanical properties and microstructure of parts, thereby affecting the overall performance and reliability of the final product. Traditional machine - learning methods require a large amount of labeled data to identify defects, but in the actual manufacturing environment, obtaining accurately labeled data is both difficult and expensive, resulting in a scarcity of available labeled data. ### Solution To solve this problem, this research proposes a self - supervised learning - based framework, using Vision Transformer (ViT) and Masked Autoencoder (MAE) models, to achieve effective detection of molten pool anomalies without the need for a large amount of labeled data. The specific steps are as follows: 1. **Self - supervised Learning Stage**: - Use a large number of unlabeled molten pool images to fine - tune the pre - trained MAE model. - MAE learns rich spatial feature representations from partially visible image blocks through the masked autoencoder method. 2. **Supervised Classification Stage**: - Transfer the fine - tuned MAE encoder parameters to two different classifiers for supervised training: - **ViT Classifier**: Directly use the fine - tuned MAE encoder parameters to initialize the ViT classifier. - **MLP Classifier**: Combine the spatial features learned by the fine - tuned MAE encoder and use a multi - layer perceptron (MLP) to perform the classification task. ### Experimental Results This method achieves an overall accuracy of 95.44% to 99.17% on limited labeled data, and the average F1 score exceeds 80%, indicating that this framework has good scalability and cost - effectiveness, and can effectively use a small amount of labeled data to achieve high - precision defect detection in the actual production environment. ### Formula Summary - The **Mean Squared Error (MSE) loss function** is used for loss calculation in the reconstruction task: \[ \text{MSE}=\frac{1}{N}\sum_{i = 1}^{N}(\text{predicted}_i-\text{target}_i)^2 \] where \(N\) is the total number of samples, \(\text{predicted}_i\) is the predicted value of the \(i\)-th sample, and \(\text{target}_i\) is the actual value of the \(i\)-th sample. Through this method, the research not only improves the accuracy of defect detection, but also reduces the dependence on a large amount of labeled data, providing an effective solution for automated quality control in the actual manufacturing environment.

In-Situ Melt Pool Characterization via Thermal Imaging for Defect Detection in Directed Energy Deposition Using Vision Transformers

In-situ defect detection in laser-directed energy deposition with machine learning and multi-sensor fusion

Multisensor fusion-based digital twin for localized quality prediction in robotic laser-directed energy deposition

Classification of Melt Pool States for Defect Detection in Laser Directed Energy Deposition Using FixConvNeXt Model

Deep Learning for Melt Pool Depth Contour Prediction From Surface Thermal Images via Vision Transformers

A Deep Learning approach to Defect Detection in Additive Manufacturing of Titanium Alloys

ETDNet: Efficient Transformer-Based Detection Network for Surface Defect Detection

Detecting and classifying hidden defects in additively manufactured parts using deep learning and X-ray computed tomography

Real-time monitoring and quality assurance for laser-based directed energy deposition: integrating co-axial imaging and self-supervised deep learning framework

Quality Monitoring in Additive Manufacturing Using Emission Spectroscopy and Unsupervised Deep Learning

Online melt pool depth estimation during directed energy deposition using coaxial infrared camera, laser line scanner, and artificial neural network

A New Method for Automatic Detection of Defects in Selective Laser Melting Based on Machine Vision

Defect detection by multi-axis infrared process monitoring of laser beam directed energy deposition

Automated Detection of Defects on Metal Surfaces using Vision Transformers

High-Performance Defect Detection Methods for Real-Time Monitoring of Ceramic Additive Manufacturing Process Based on Small-Scale Datasets

A Deep Learning Based Automatic Defect Analysis Framework for In-situ TEM Ion Irradiations

In-situ surface porosity prediction in DED (directed energy deposition) printed SS316L parts using multimodal sensor fusion

Research on Automated Defect Classification Based on Visual Sensing and Convolutional Neural Network-Support Vector Machine for GTA-Assisted Droplet Deposition Manufacturing Process

A Real-Time Defect Detection Strategy for Additive Manufacturing Processes Based on Deep Learning and Machine Vision Technologies

Automated Defect Recognition for Additive Manufactured Parts Using Machine Perception and Visual Saliency

Rapid surface defect identification for additive manufacturing with in-situ point cloud processing and machine learning