Digital Twin of Wire Arc Additive Manufacturing

Wire Arc Additive Manufacturing (WAAM) is an emerging manufacturing technology that has been used to build medium to larger-sized components. With the rise of Industry 4.0 and smart manufacturing, production automation, intelligence, and digitalization has attracted increasing scientific attention in manufacturing field. Despite the adoption of intelligent algorithms in WAAM, their application remains fragmented, targeting specific issues without a unified, intelligent optimization system. This thesis aims to bridge this gap by introducing a comprehensive Digital Twin (DT) framework that integrates monitoring, modelling, control, and simulation to enhance WAAM processes.

In this study, the monitoring system of WAAM-DT framework consists of a high-frequency bead geometry measuring system and a sensor-fusion defect detection system. Utilizing electric signal sensors, the system captures welding parameters to predict bead dimensions with high accuracy, leveraging Multi-layer Perceptron (MLP) regressors and an incremental learning framework for optimized performance. Comparative studies underscore the system's efficacy, offering speed and precision superior to traditional methods.

Defect detection system includes multi-scale sensing technologies: a welding electric signal sensor, a camera, and a laser profilometer are used to collect welding current and voltage data, image data, and point cloud data. The collected multi-scaled data are subsequently analysed by Multi-layer Perceptron (MLP) classifier and YOLOv5 in temporal and spatial scale, respectively, then fused in decision-level via a Variational Autoencoder (VAE). The system performance is then tested to detect defects and geometric errors in practical experiments and the results show that the overall F1 score is 0.791, including detecting, classifying, and analysing the cause of defects. Additionally, the total predicting time is within 0.5 s, which is suitable for an in-process monitoring system.

The WAAM-DT framework also introduces an adaptive simulation model for predicting physical distortions during deposition, using a novel diffusion model architecture for spatial and temporal analysis. Pretrained offline with Finite Element Method (FEM) simulated distortion fields, the model successfully predicts distortion fields online using laser-scanned distortion fields during the deposition process. Experimental validation on seven thin-wall structures demonstrated its superior performance, achieving a Root Mean Square Error (RMSE) below 0.9 m, outperforming FEM by 143% and Artificial Neural Networks (ANN) based methods by 151%, marking a significant stride towards realizing an WAAM-DT.

Furthermore, an innovative control system combining a linear autoregressive (ARX) model and a multi-input multi-output (MIMO) model-predictive control (MPC) algorithm optimizes geometric outcomes of WAAM processes in real-time. Through both simulation and experiments, results show that the real-time control performance is improved by increasing the complexity of implemented control algorithm: Controlled geometric fluctuations in the test component were reduced by 200% whilst maintaining fluctuations within a 3 mm limit under various welding conditions. In addition, the adaptiveness of designed control strategy is verified by accurately controlling the fabrication of a part with complex geometry.

This thesis lays the groundwork for a robust WAAM-DT system, presenting a holistic approach to addressing the technology's current limitations. Future work focus on investigation and modelling of the deposition mechanism to further refining the WAAM-DT's predictive accuracy and operational efficiency.