Machine learning device, additive manufacturing system, machine learning method for welding condition, method for determining welding condition, and a non-transitory computer readable medium

An additive manufacturing system for manufacturing an additively-manufactured object by welding a filler metal and depositing weld beads, the additive manufacturing system comprising:

• at least one hardware processor configured to
• create, based on design data of the additively-manufactured object, a plurality of pieces of pass data corresponding to a plurality of weld beads forming the additively-manufactured object;
• determine welding conditions for forming weld beads corresponding to the respective pieces of created pass data;
• specify one or more block patterns formed by the weld beads corresponding to the pass data based on an arrangement when the weld bead is formed;
• derive shape data corresponding to pass data by inputting a determined welding condition corresponding to the pass data and the specified block pattern to a learned model generated by performing a learning process using a welding condition of a weld bead and a block pattern formed by the weld bead as input data and shape data of the weld bead as output data, and derive the shape data corresponding to the pass data in order, for the specified one or more block patterns, based on a priority defined in advance for block patterns;
• and
• adjust the welding condition such that a difference between the derived shape data and shape data indicated by the created pass data does not exceed a predetermined threshold.