Machine learning device
Provided is a machine learning device capable of causing idling of a tool, and of calculating oscillation conditions that realize favorable evaluation data of finished workpieces. The machine learning device that learns oscillation conditions of a machine tool that performs oscillation cutting while oscillating a tool and a workpiece relative to each other is provided with a set conditions acquisition unit that acquires set conditions for the oscillation cutting, a label acquisition unit that acquires evaluation data of finished workpieces by the machine tool as a label, and a learning unit that performs supervised learning using a set of the set conditions and the label as training data, the learning unit being provided with a learning model for training oscillation conditions that optimize the evaluation data of the finished workpieces.
1 . A computer device for learning an oscillation condition for a machine tool that performs oscillation machining while oscillating a tool and a workpiece relative to each other, comprising:
a processor,
wherein the processor is configured to:
acquire a condition setting for the oscillation machining;
acquire as a label, evaluation data on the workpiece machined with the machine tool, the evaluation data on the machined workpiece including data on a surface roughness, roundness, or dimensional accuracy of the machined workpiece; and
perform supervised learning using a pair of the condition setting and the label as teacher data,
wherein the processor includes a learning model for learning an oscillation condition where the evaluation data on the machined workpiece is optimized to minimize the surface roughness of the machined workpiece, minimize the roundness of the machined workpiece, or maximize a dimensional accuracy of the machined workpiece.
2 . The computer device according to claim 1 , the processor outputs, based on the learning model, an optimal oscillation condition where the evaluation data on the machined workpiece is optimized.
3 . The computer device according to claim 2 , wherein
the processor calculates a chip shredding oscillation condition where a chip from the workpiece can be shredded, and
the processor outputs, based on the learning model, the optimal oscillation condition satisfying the chip shredding oscillation condition.
4 . The computer device according to claim 2 , wherein
the processor calculates an upper limit oscillation condition not exceeding a preset upper limit, and
the processor outputs, based on the learning model, the optimal oscillation condition satisfying the upper limit oscillation condition.
5 . The computer device according to claim 1 , wherein
the condition setting includes a machining condition including at least one selected from a tool feed speed, a rotation speed about a main axis, a coordinate value, a tool blade edge, or a tool type in the machine tool and a workpiece material, and
the oscillation condition including at least one selected from an oscillation frequency or an oscillation amplitude of the machine tool.
6 . The computer device according to claim 1 , wherein the computer device is shared by a plurality of numerical control devices.
7 . The computer device according to claim 1 , wherein the computer device is provided on a cloud server.
8 . A compute device for learning an oscillation condition for a machine tool that performs oscillation machining while oscillating a tool and a workpiece relative to each other, comprising:
a processor,
wherein the processor is configured to:
acquire as condition information, a condition setting for the oscillation machining;
acquire as determination information, evaluation data on the workpiece machined with the machine tool, the evaluation data on the machined workpiece including data on a surface roughness, roundness, or dimensional accuracy of the machined workpiece;
output action information indicating how to change the oscillation condition from a current condition;
calculate a value of a reward in reinforcement learning based on the determination information;
update a value function for setting a value of the oscillation condition for the machine tool based on the condition information, the action information, and the reward;
and output based on the value function, an optimal oscillation condition where the evaluation data on the machined workpiece is optimized to minimize the surface roughness of the machined workpiece, minimize the roundness of the machined workpiece, or maximize a dimensional accuracy of the machined workpiece.
9 . The computer device according to claim 8 , wherein the processor calculates the reward as a positive value when the evaluation data on the machined workpiece is less than a predetermined threshold, and calculates the reward as a negative value when the evaluation data exceeds the predetermined threshold.
10 . The computer device according to claim 9 , wherein in a case where the evaluation data is the surface roughness on the machined workpiece, the processor determines the predetermined threshold based on a theoretical surface roughness approximate expression.
11 . The computer device according to claim 10 , wherein the surface roughness includes at least one selected from an arithmetic average roughness, a maximum height, a maximum peak height, a maximum valley depth, an average height, a maximum peak-to-valley height, or a load length ratio.
12 . The computer device according to claim 8 , wherein
the processor calculates a chip shredding oscillation condition where a chip from the workpiece can be shredded, and
the processor outputs the optimal oscillation condition satisfying the chip shredding oscillation condition.
13 . The computer device according to claim 8 , wherein
the processor calculates an upper limit oscillation condition not exceeding a preset upper limit, and
the processor outputs the optimal oscillation condition satisfying the upper limit oscillation condition.