Machine learning device, laser machine, and laser machining system

A machine learning device that performs machine learning on a laser machine comprising a plurality of galvanometer mirrors for reflection of a laser beam and a plurality of galvanometer motors for driving and rotating corresponding ones of the galvanometer mirrors according to driving control data generated based on a position command indicating a machining target position on a workpiece, and scanning the laser beam over the workpiece, the machine learning device comprising:

• input data acquisition unit that acquires, as input data, a detected temperature of at least one galvanometer mirror of the plurality of galvanometer mirrors and a detected temperature of at least one galvanometer motor of the plurality of galvanometer motors that drives and rotates the at least one galvanometer mirror;
• label acquisition unit that acquires, as a label, a coefficient for calculating the machining target position from an actual position of machining with the laser beam on the workpiece relative to the machining target position, the coefficient being acquired from a coefficient calculation unit before calculation of the machining target position and the actual machining position, the coefficient calculation unit acquiring machining target position data at the time of acquisition of the detected temperature of the at least one galvanometer mirror and the detected temperature of the at least one galvanometer motor as the input data by the input data acquisition unit, and concurrently obtaining the coefficient by solving a plurality of mathematical expressions;
• and
• learning unit that performs supervised learning using a set of the label and the input data as training data to construct a mathematical model for calculating the machining target position from the actual machining position on the workpiece on the basis of the at least two detected temperatures consisting of the detected temperature of the at least one galvanometer mirror and the detected temperature of the at least one galvanometer motor, the learning unit calculating a plurality of coefficients defined based on the at least two detected temperatures consisting of the detected temperature of the at least one galvanometer mirror and the detected temperature of the at least one galvanometer motor according to the mathematical model using the training data, and outputting the mathematical model.