Federated learning for additive manufacturing and quality enhancement

Overview

This project aligns with the megatrend of digitalization, which is reshaping technology, the economy, and society through profound IT-driven changes. Reflecting the focus of the "Digital Transformation" research cluster at Hochschule Bremen, it integrates technological foundations in data processing and computer science with usercentered solutions in interdisciplinary applications. Specifically, the project aims to develop a federated learning (FL) framework for improving anomaly detection in additive manufacturing (AM) processes. Enabling collaborative model training across diverse manufacturing environments without sharing sensitive data enhances defect prediction and quality control. The framework incorporates techniques like feature augmentation to address diverse data distributions and advanced methods for real-time anomaly detection. Its performance will be evaluated using metrics such as accuracy and F1 score, comparing FL with centralized models to assess improvements in generalization and privacy. Furthermore, privacy-preserving approaches like differential privacy and secure computation will ensure data confidentiality, contributing to the broader societal and industrial goals of safe and efficient digital transformation.