Life estimation device and machine learning device
AssigneeFanuc Corporation
InventorZhizhen Xia
Life of this patent
Estimating the term…
Abstract
A device that estimates a life of a clamping mechanism clamping rotation of a rotary table includes a machine learning device. The machine learning device observes operating state data of the rotary table and operation history data of the rotary table as a state variable indicative of a current state of an environment, and acquires life data indicative of the life of the clamping mechanism as label data. In addition, the device uses the state variable that has been observed and the label data that has been acquired and learns the operating state data and the operation history data and the life data in association with each other.
Claims7
A life estimation device for estimating a life of a clamping mechanism for clamping or unclamping rotation of a rotary table, wherein the life of the clamping mechanism is a period of a time until maintenance of the clamping mechanism becomes necessary, the life estimation device comprising:
- a machine learning device configured to learn estimation of the life of the clamping mechanism with respect to a state of the rotary table, the machine learning device comprising a processor configured to
- observe operating state data regarding an operating state of the rotary table and operation history data regarding an operation history of the rotary table as a state variable indicative of a current state of an environment,
- acquire, as label data, life data indicative of the life of the clamping mechanism,
- learn a correlation between the operating state data and the operation history data and the life data using the state variable and the label data based on a multilayered structure in accordance with a neural network using the state variable as an input,
- output a result of estimation of the life of the clamping mechanism on a basis of a result of learning, and
- update a correlation model for estimating the life of the clamping mechanism,
- wherein the operating state data include weights of a jig and a workpiece attached to the rotary table, and
- the operation history data include a cumulative operation time of the rotary table.
The life estimation device according to 1, wherein
- the machine learning device resides in a cloud server.
A machine learning device for learning estimation of a life of a clamping mechanism for clamping or unclamping rotation of a rotary table with respect to a state of the rotary table, wherein the life of the clamping mechanism is a period of a time until maintenance of the clamping mechanism becomes necessary, the machine learning device comprising:
- a processor configured to
- observe operating state data regarding an operating state of the rotary table and operation history data regarding an operation history of the rotary table as a state variable indicative of a current state of an environment,
- acquire, as label data, life data indicative of the life of the clamping mechanism, the life data being acquired, and
- learn a correlation between the operating state data and the operation history data and the life data using the state variable and the label data based on a multilayered structure in accordance with a neural network using the state variable as an input,
- output a result of estimation of the life of the clamping mechanism on a basis of a result of learning, and
- update a correlation model for estimating the life of the clamping mechanism,
- wherein the operating state data include weights of a jig and a workpiece attached to the rotary table, and
- the operation history data include a cumulative operation time of the rotary table.
Dependencies are inferred from the claim text.