A New Method for Automatic Detection of Defects in Selective Laser Melting Based on Machine Vision

Area of Science:

• Additive Manufacturing
• Materials Science
• Quality Control

Background:

• Selective Laser Melting (SLM) is a key metal additive manufacturing technology.
• Current monitoring methods focus on the melt pool, limiting real-time quality control.
• Powder spreading quality is critical for the final part integrity in SLM.

Purpose of the Study:

• To develop an effective method for detecting powder spreading defects in SLM.
• To improve real-time quality control of SLM processes.
• To enhance the overall quality of additively manufactured parts.

Main Methods:

• Utilizing industrial cameras to capture images of the powder spreading surface.
• Developing image processing algorithms to identify common powder spreading defects.
• Employing machine learning classifiers, specifically Multilayer Perceptron (MLP), for defect identification.

Main Results:

• Successfully extracted and classified three common powder spreading defects.
• The Multilayer Perceptron (MLP) classifier demonstrated the highest accuracy.
• The detection method achieved a high recognition rate and fast detection speed.

Conclusions:

• The proposed defect detection method is efficient and accurate for SLM processes.
• This approach enables real-time feedback control to improve part quality.
• The method supports the high forming efficiency required in additive manufacturing.