Penetration Depth Modeling and Process Parameter Maps for Laser Welds Using Machine Learning

Penetration control is an important factor in determining the weld quality keyhole mode laser welding, which enables deep penetration. In this study, machine learning models and neural network were developed by using 380 published welding data constructed for steel base metals under following conditions: a power of 0.3-16.7 kW, speed 0.3-20.0 m/min, bead diameter 0.05-0.78 mm. A model SVM (supported vector machine) could accurately predict penetration depth with coefficient determination, R2 0.95. shallow five nodes only one hidden layer was slightly improved accuracy 0.98. It confirmed that neither overfitted, process parameters (welding beam diameter) maps contours provided 2-8 kW.