To be able to produce first-time right and zero-defect parts, in-process monitoring and control is essential. Previous projects (Monicon, ViL) already indicate that a single optical sensor signal (photodiode, camera) can successfully be correlated with defect formation for the defects studied in those projects (mainly keyhole porosity) but does not give enough information to know exactly when and where a defect is forming.
This project therefore aims to use multiple sensors of a distinctly different type (existing optical sensors, acoustic emission sensors mounted on the build module, and eddy current sensors mounted on the LPBF recoater), and further combines them with virtual sensing, online data-analysis, machine-learning assisted sensor fusion, and in-process control possibilities (through process parameter changes) to allow timely corrective action. Given the complexity of combining different monitoring data streams, the project will employ self-learning models. These models will be trained and validated by compiling an extensive library of process parameters, associated monitoring signals, and part properties as measured by X-ray CT (XCT) and the monitoring know-how previously generated in Monicon, ViL, and other internal projects.
The application field in MuSIC_SBO is extended from LPBF to welding-based manufacturing processes in general (also WAAM, DED, …) and from only keyhole pores to different defect types and processing regimes (lack of fusion pores, spatter, cracks, conduction vs keyhole melting). While the models probably need to be adapted to the specific process that is being studied, the project aims to develop general-purpose sensor data collection, structuring, and processing algorithms to enable a multiple loop control system that can be used for many thermally driven manufacturing processes.
Two control loops are envisioned:
- One real-time loop that can act based on sensors like those used in Monicon and ViL, which should already reduce the tendency for defect formation.
- A second layer-by-layer control system can be used for corrective action using fused sensor metrics that cannot be calculated in real time, or the eddy current signal, which is considered the first true in situ quality control measurement rather than process monitoring.
MuSIC_SBO is a Strategic Basic Research SBO project. We are looking for companies to join the User Group and work with us on the valorisation of the project.
Interested? Complete the form below and we will contact you as soon as possible.