AI-Based Multi-Sensor Quality Assurance in Additive Manufacturing

In our project at TH Rosenheim, we focus on developing an innovative AI-driven method for quality control in Laser Powder-Bed Fusion (LPBF) metal Additive Manufacturing (AM). By integrating multi-sensor data, defects can be identified in real time during the 3D printing process. This is crucial for sectors such as aerospace, automotive, and medical devices, where component reliability is non-negotiable.

Laser etching view inside the printing chamber (Source: TH Rosenheim, CC BY, HKE).

Multi-Sensor Detection and AI Integration

We propose a system that combines visual monitoring with eddy current sensors. Leveraging deep learning, the system merges data from various sensors to enable early defect detection. This allows immediate intervention, reducing energy consumption, material waste, and production time. Automated classification of defect types further enhances efficiency and reliability.

Dashboard for live sensor data monitoring during the LPBF process (Source: TH Rosenheim, CC BY, HKE).

Technical Challenges and Solutions

Integrating multiple sensors requires precise data synchronization and acquisition strategies. We address these challenges with a robust framework that ensures accurate alignment between sensor inputs. The collected data not only supports real-time monitoring but also provides documentation for process validation and knowledge base development.

Defect detection results used to refine and improve the AM process (Source: TH Rosenheim, CC BY, HKE).

Towards Smarter Additive Manufacturing

Our AI-based approach improves the precision of non-destructive testing (NDT) compared to traditional post-production inspections. By enhancing in-situ monitoring, industries benefit from more reliable process control, reduced failure rates, and sustainable production practices.