Tool state estimation apparatus and machine tool
A scattering direction, a scattering range, a scattering speed, or the like of chip generated by machining is acquired by image pickup unit, and a state (such as wearing and fracture) of a tool is estimated from the acquired dynamic information. A machine learning device is used to perform the estimation and the machine learning device is caused to learn a state of a tool relative to the combination of a plurality of dynamic information on the vicinity of the tool, and the state of the tool is estimated based on a result of the learning.
1. A tool state estimation apparatus for estimating a state of a tool used by a machine tool to machine a workpiece, the tool state estimation apparatus comprising:
a processor configured to:
acquire dynamic information on the vicinity of the tool from log data acquired during an operation of the machine tool,
generate input data based on the acquired dynamic information on the vicinity of the tool, the dynamic information including dynamic information that includes at least scattering state of chips in the machining of the workpiece, and
perform machine learning, using the generated input data, to construct a learning model; and
a storage configured to store the constructed learning model.
2. The tool state estimation apparatus according to claim 1 , wherein
the machine learning is supervised learning, and
the processor is further configured to
acquire, from the log data acquired during the operation of the machine tool, teacher data indicating the state of the tool corresponding to the generated input data, and
perform the supervised learning, using the generated input data and the acquired teacher data, to construct the learning model.
3. The tool state estimation apparatus according to claim 1 , wherein
the machine learning is unsupervised learning, and
the processor is configured to perform the unsupervised learning, using the generated input data, to construct the learning model in which a cluster based on the dynamic information on the vicinity of the tool is generated.
4. The tool state estimation apparatus according to claim 1 , wherein
the dynamic information on the vicinity of the tool is information on an image indicating a scattering state of chips generated by machining of the machine tool.
5. The tool state estimation apparatus according to claim 4 , wherein the image is acquired during the operation of the machine tool.
6. The tool state estimation apparatus according to claim 1 , wherein
the processor is further configured to
acquire static information on the machining from the log data acquired during the operation of the machine tool, and
generate the input data based on the acquired static information as well as the dynamic information on the vicinity of the tool.
7. The tool state estimation apparatus according to claim 1 , wherein
the scattering state includes at least one of (1) central angle of scattering range of the chips, (2) scattering distance of the chips, or (3) scattering speed of the chips.
8. A tool state estimation apparatus for estimating a service life of a tool used by a machine tool to machine a workpiece, the tool state estimation apparatus comprising:
a storage configured to store a learning model constructed by machine learning based on dynamic information on the vicinity of the tool acquired during an operation of the machine tool, the dynamic information including dynamic information that includes at least scattering state of chips in the machining of the workpiece; and
a processor configured to:
acquire dynamic information on the vicinity of the tool from log data acquired during the operation of the machine tool,
generate input data based on the acquired dynamic information on the vicinity of the tool, the dynamic information including dynamic information that includes at least scattering state of chips in the machining of the workpiece, and
estimate, using the learning model, a state of the tool from the generated input data.
9. The tool state estimation apparatus according to claim 8 , wherein
the learning model is constructed by the machine learning based on both (i) the dynamic information on the vicinity of the tool acquired during the operation of the machine tool and (ii) static information on the machining, and
the processor is further configured to
acquire static information on the machining from the log data acquired during the operation of the machine tool, and
generate the input data based on the acquired static information as well as the dynamic information on the vicinity of the tool.
10. A machine tool, comprising:
an input/output section communicating with a tool state estimation apparatus which is configured to estimate a service life of a tool used by the machine tool to machine a workpiece, the tool state estimation apparatus comprising:
a storage configured to store a learning model constructed by machine learning based on dynamic information on the vicinity of the tool acquired during an operation of the machine tool, the dynamic information including dynamic information that includes at least scattering state of chips in the machining of the workpiece; and
a processor configured to:
acquire dynamic information on the vicinity of the tool from log data acquired during the operation of the machine tool,
generate input data based on the acquired dynamic information on the vicinity of the tool, the dynamic information including dynamic information that includes at least scattering state of chips in the machining of the workpiece, and
estimate, using the learning model, a state of the tool from the generated input data; and
an alert section configured to issue an alert based on an estimation result of the state of the tool by the processor of the tool state estimation apparatus.