Preliminary study on droplet re-aggregation forming and active control of weld morphology based on programmable metal transfer

On the basis of double-sided dual pulse laser-driven programmable metal transfer in gas metal arc welding, a back propagation neural network prediction model was established with dual laser power and pulse width as inputs and molten droplet deflection angle as an output. The reliability of the model was verified through process experiments. Preliminary exploration was conducted on the control effect of droplet re-aggregation under the action of dual pulse laser, as well as the active creation of weld morphology. The results indicate that when the base material melts more and forms a high-temperature melt pool, the droplets entering the melt pool according to the set trajectory can actively regulate the weld morphology to a certain extent. When the base material melts very little, a programmable droplet path can form a droplet assembly manufacturing similar to metal droplet spray deposition. By matching the process parameters of double-sided laser pulses, the formation of “two welds in one direction” and “wave shaped” weld formation, as well as the distribution of “zigzag shaped” and “semicircular” molten droplets, were obtained, and the assembly type weld seam formation and molten droplet formation were preliminarily achieved.