Machine learning device, cleaning prediction device, and cleaning system
A machine learning device has an input data acquisition unit that acquires input data including an arbitrary machining condition for an arbitrary workpiece to be machined by an arbitrary machine tool, a cleaning condition for cleaning, by an arbitrary cleaning device, the inside of a machine by which the workpiece was machined under the machining condition and state information indicating a degree of dirtiness of the inside of the machine before the machining under the machining condition. The device also has a label acquisition unit that acquires, as label data, state information about the inside of the machine after machining under the machining condition and cleaning under the cleaning condition included in the input data, and a learning unit that executes supervised learning using the input data and the label data, so as to generate a learned model.
1 . A machine learning apparatus comprising:
a memory configured to store a program; and
a processor configured to execute the program and control the machine learning apparatus to:
acquire input data, the input data including arbitrary machining conditions for an arbitrary machine tool to machine arbitrary workpieces, washing conditions for an arbitrary washing device to wash an inside of a machine tool where the arbitrary workpieces have been machined under the arbitrary machining conditions, and state information indicating a degree of soil inside the machine tool before machining under the arbitrary machining conditions;
acquire, as label data, state information about the inside of the machine tool after performing machining under the arbitrary machining conditions included in the input data and subsequent washing under the washing conditions included in the input data; and
execute supervised learning using the acquired input data and the acquired label data, and generate a trained model.
2 . The machine learning apparatus according to claim 1 , wherein
the machining conditions include at least a type of the machine tool, material of workpieces, a type of a cutting tool, a spindle speed, a feed rate, a depth of cut, and a quantity of workpiece production indicating a number of workpieces to be machined; and
the washing conditions include at least number of times of predetermined washing.
3 . A washing prediction apparatus comprising:
a second memory configured to store a second program and the trained model generated by the machine learning apparatus according to claim 1 ; and
a second processor configured to execute the second program and control the washing prediction apparatus to:
receive, prior to machining by the machine tool, inputs of machining conditions to be applied, washing conditions for predetermined washing, and current state information about an inside of the machine tool;
input, to the trained model, the received inputs of the machining conditions to be applied, the washing conditions and the current state information about the inside of the machine tool, and predict state information about the inside of the machine tool after performing machining under the machining conditions to be applied and subsequent washing under the washing conditions;
decide a washing timing at which a quantity of workpiece production is largest, based on a comparison between a predicted value included in the state information about the inside of the machine tool and a preset threshold; and
output the decided washing timing to at least the machine tool,
wherein the inside of the machine tool is washed based on the washing timing output from the second processor.
4 . A washing prediction apparatus comprising;
a second memory configured to store a second program and the trained model generated by the machine learning apparatus according to claim 1 ; and
a second processor configured to execute the second program and control the washing prediction apparatus to:
receive, prior to machining by the machine tool, inputs of machining conditions to be applied, washing conditions for predetermined washing, and current state information about an inside of the machine tool;
input, to the trained model, the received inputs of the machining conditions to be applied, the washing conditions and the current state information about the inside of the machine tool, and predict state information about the inside of the machine tool after performing machining under the machining conditions to be applied and subsequent washing under the washing conditions;
decide a smallest number of times of the predetermined washing, for washing of the inside of the machine tool after performing the machining under the machining conditions to be applied, based on a comparison between a predicted value included in the state information about the inside of the machine tool predicted by the prediction unit and a preset threshold; and
output the decided washing timing to at least the machine tool,
wherein the inside of the machine tool is washed based on the washing timing output from the second processor.
5 . The washing prediction apparatus according to claim 4 , wherein content of the predetermined washing is changed according to the smallest number of times of the predetermined washing which was decided.
6 . The washing prediction apparatus according to claim 3 , wherein the trained model is provided in a server connected so as to be accessible from the washing prediction apparatus via a network.
7 . The washing prediction apparatus according to claim 3 , comprising
the machine learning apparatus.
8 . A washing system comprising:
the washing prediction apparatus according to claim 3 ; and
a washing device.
9 . The washing system according to claim 8 , wherein the washing device is provided inside the machine tool.
10 . The washing system according to claim 8 , wherein the washing device is attached to or included in a robot.