Failure prediction method and failure prediction apparatus
AssigneeSeiko Epson Corporation
InventorsToshinori Hiraide, Yimei Ding
Life of this patent
Estimating the term…
Abstract
A failure prediction method of predicting a failure of a component of a robot including a robot arm having the component and a detection section that detects information on vibration characteristics when the robot arm moves, includes generating a failure prediction model for prediction of the failure of the component by machine learning based on the information on vibration characteristics, and predicting the failure of the component based on an estimated value of failure prediction output by the generated failure prediction model when the information on vibration characteristics is input to the generated failure prediction model.
Claims7
A failure prediction method for causing a processor to execute a process to predict a failure of a reducer of a joint of a robot arm of a robot, the failure prediction method comprising executing on the processor the steps of:
- controlling a motor to repeatedly move the robot arm several times to obtain information on vibration characteristics of the robot arm due to movement of the reducer, the information on the vibration characteristics including detection results of an encoder at the joint, detection results of an inertial sensor attached at a distal end of the robot arm, detection results of a force detection sensor disposed at the robot arm, and detection results of a sound collection sensor;
- generating a failure prediction model for prediction of the failure of the reducer by machine learning based on the information on the vibration characteristics;
- and
- predicting the failure of the reducer based on an estimated value of failure prediction output by the generated failure prediction model when the information on the vibration characteristics is input to the generated failure prediction model,
- wherein, when a degree of the vibration characteristics obtained at a current time is equal to or more than a predetermined value and is more than a degree of the vibration characteristics obtained at a previous time, the processor is configured to predict an occurrence of the failure of the reducer.
A failure prediction apparatus comprising:
- a robot, the robot having:
- a robot arm;
- and
- a joint in the robot arm, the joint configured to move the robot arm and having a motor and a reducer;
- and
- a processor configured to be programed to:
- control the motor to repeatedly move the robot arm several times to obtain information on vibration characteristics of the robot arm due to movement of the reducer, the information on the vibration characteristics including detection results of an encoder at the joint, detection results of an inertial sensor attached at a distal end of the robot arm, detection results of a force detection sensor disposed at the robot arm, and detection results of a sound collection sensor;
- generate a failure prediction model for prediction of the failure of the reducer by machine learning based on the information on the vibration characteristics;
- and
- predict the failure of the reducer based on an estimated value of failure prediction output by the generated failure prediction model when the information on the vibration characteristics is input to the generated failure prediction model,
- wherein, when a degree of the vibration characteristics obtained at a current time is equal to or more than a predetermined value and is more than a degree of the vibration characteristics obtained at a previous time, the processor is configured to predict an occurrence of the failure of the reducer.
Dependencies are inferred from the claim text.