THERMAL DISPLACEMENT CORRECTION DEVICE

To provide a thermal displacement correction device capable of efficiently performing temperature correction of machine accuracy when a peripheral situation is an unknown environment.SOLUTION: A machine learning device 100 provided in a thermal displacement correction device 1 includes: a model data selection section 130 which selects a learned model suitable for additional learning of thermal displacement correction amount of each shaft provided in a manufacturing machine 70 for an operation state of the manufacturing machine under a change of a peripheral temperature of the manufacturing machine 70; and a learned model storage section 200 which stores a pattern of a peripheral temperature change curve indicating the transition of the change of the peripheral temperature of the manufacturing machine 70 and the learned model learned under the change of the peripheral temperature in association with each other. The model data selection section 130 selects the learned model suitable for additional learning of the thermal displacement correction amount of each shaft provided in the manufacturing machine 70 for the operation state of the manufacturing machine 70 on the basis of the peripheral temperature change curve stored in the learned model storage section 200.SELECTED DRAWING: Figure 2 【課題】周辺状況が未知の環境における機械精度の温度補正を効率的に行うことを可能とする熱変位補正装置を提供すること。【解決手段】本発明の熱変位補正装置１が備える機械学習装置１００は、製造機械７０の周辺温度の変化の下で、製造機械７０の動作状態に対する該製造機械が備える各軸の熱変位補正量を追加学習するのに適した学習済みモデルを選択するモデルデータ選択部１３０と、製造機械７０の周辺温度の変化の推移を示す周辺温度変化曲線のパターンと、該周辺温度の変化の下で学習された学習済みモデルとを関連付けて記憶する学習済みモデル記憶部２００と、を備え、モデルデータ選択部１３０は、学習済みモデル記憶部２００に記憶された周辺温度変化曲線に基づいて、製造機械７０の動作状態に対する該製造機械７０が備える各軸の熱変位補正量を追加学習するのに適した学習済みモデルを選択する。【選択図】図２