Machine learning device for machine tool and thermal displacement compensation device

A machine learning device that optimizes an equation for estimating a thermal displacement amount of a machine element that thermally expands by way of machine learning, based on operating state data representing an operating state of the machine element of a machine tool having the machine element, the machine learning device comprising:

• a data acquisition means for acquiring the operating state data of the machine element;
• a thermal displacement amount acquisition means for acquiring measured values for a thermal displacement amount of the machine element;
• a storage means for storing the operating state data of the machine element acquired by the data acquisition means and measured values for the thermal displacement amount of the machine element acquired by the thermal displacement amount acquisition means as training data associated with each other as a label;
• an equation setting means for setting an equation for calculating the thermal displacement amount of the machine element, by performing machine learning based on the operating state data of the machine element and the measured values for the thermal displacement amount of the machine element;
• a thermal displacement amount calculation means for calculating an estimated value for the thermal displacement amount of the machine element, by substituting operating state data of the machine element within a predetermined time period stored as training data in the storage means, into the equation set by the equation setting means;
• and
• a determination means for determining whether a difference between the estimated value for the thermal displacement amount of the machine element within the predetermined time period calculated by the thermal displacement amount calculation means and the measured value for the thermal displacement amount of the machine element within the predetermined time period stored as training data in the storage means is no more than a predetermined threshold,
• wherein the equation setting means sets again the equation in a case of being determined that the difference is not no more than a predetermined threshold by way of the determination means, and sets said equation as an optimal equation in a case of being determined that the difference is no more than a predetermined threshold by way of the determination means.