Image-based algorithm for nozzle adhesion detection in powder-fed directed-energy deposition

• The rapidly growing technological innovation of directed energy deposition leads to an increase in part complexity as well as quality and mechanical properties of manufacturable components. However, the variety of process parameters and influencing factors still requires skilled operators, who observe the machine tools. For an unobserved use of deposition welding machines, well parametrized and validated monitoring systems have to analyze the process to detect irregularities and finally initiate a machine stop. This study focuses on nozzle adhesions that frequently occur when tool or high-speed steels are processed. This effect leads to decreasing quality or ultimately to a failure of the whole welding process. In this work, the authors present an algorithm and the corresponding parametrization to automatically detect nozzle adhesions based on images from a coaxial camera, integrated in the laser head. The algorithm is based on a detailed image analysis from which temporal and spatial patterns are derived. In particular, the algorithm calculates a nozzle adhesion indicator based on the heat intensity distribution in an experimentally derived shaped area on the inner nozzle boundary. It is parametrized in such a way that process-critical adhesions are detected. The algorithm was parametrized using an experimental setup with four materials: stainless steel (X2CrNiMo17-12-2), tool steel (X35CrMoMn7-2-1), high-speed steel (HS6-5-2C), and the nickel-based alloy NiCr19NbMo.

Subject headings

TEKNIK OCH TEKNOLOGIER  -- Materialteknik -- Bearbetnings-, yt- och fogningsteknik (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Materials Engineering -- Manufacturing, Surface and Joining Technology (hsv//eng)

Keyword

directed energy deposition

laser metal deposition

laser cladding

nozzle adhesion

image processing

melt pool

coaxial monitoring

process monitoring

Produktionsutveckling

Manufacturing Systems Engineering

Publication and Content Type

ref (subject category)

art (subject category)