A method for backlash compensation in motion control systems includes homing a payload of a motion control system, and performing a tooth-pitch non-uniformity procedure on the motion control system to identify a non-uniformity correction. A backlash lookup table is generated for use in backlash correction during normal operation of the motion control system. The backlash lookup table is generated using a training process that includes selecting a move sequence for operating the motion control system, and executing the move sequence with the non-uniformity correction. The training process further includes computing a backlash measurement describing backlash of one or more components of the motion control system during the move sequence, and storing the backlash measurement in the backlash lookup table.

A motor control apparatus includes an error calculation unit which calculates an error between a position of a movable unit and a position of a driven unit, a memory unit which memorizes the error in association with a torque command value as an initial error, and a compensation amountcompensation amount calculation unit which calculates a compensation amountcompensation amount for compensating an elastic deformation amount of an elastic factor between the movable unit and the driven unit. The compensation amountcompensation amount calculation unit calculates the compensation amountcompensation amount based on the initial error held by the memory unit, the torque command value held in association with the initial error, and a torque command value calculated by a torque command calculation unit when a motor rotates reversely.

Stop and start detection block (63) determines whether or not a joint portion is in a stopped state before a rotation direction of the joint portion is inverted based on stop flag signal (Stop_Flg). When it is determined that the joint portion is in the stopped state, filter processing block (65) changes a frequency component of a correction amount for correcting backlash to a low frequency lower than a predetermined threshold value.

In an angular velocity calculation block, an angular velocity component is calculated based on a position command for a joint portion. In a kinetic calculation block, a kinetic torque is calculated based on the position command for the joint portion. In a command velocity component reversal detection block, a reversal timing is calculated based on the angular velocity component. In a reverse torque detection block, a reverse torque is calculated based on the kinetic torque and the reversal timing. In a backlash correction amount calculation block, the correction amount of the joint portion is calculated based on the kinetic torque, the reverse torque, and the reversal timing.

Regarding a backlash amount measurement system provided in the servo control device of the drive motor of machine tools and industrial machines, a technical problem is to accurately measure the backlash amount at the time of inversion of a drive shaft by means of a backlash amount measurement system having a simpler configuration with a small number of detectors. The above technical problem can be solved by a backlash amount measuring unit equipped with an inversion detection unit which detects the start of inversion of the shaft, an arrival detection unit which detects that the backlash end has been reached from the start of inversion, and a time/distance measurement unit which measures the time from when the inversion detection unit detects the start of inversion until the arrival detection unit detects that the backlash end has been reached, or the movement distance.

In an angular velocity calculation block, an angular velocity component is calculated based on a position command for a joint portion. In a kinetic calculation block, a kinetic torque is calculated based on the position command for the joint portion. In a command velocity component reversal detection block, a reversal timing is calculated based on the angular velocity component. In a reverse torque detection block, a reverse torque is calculated based on the kinetic torque and the reversal timing. In a backlash correction amount calculation block, the correction amount of the joint portion is calculated based on the kinetic torque, the reverse torque, and the reversal timing.

To perform reinforcement learning enabling to prevent complicated adjustment of coefficients of backlash compensation and backlash acceleration compensation. A machine learning apparatus includes a state information acquiring part for acquiring, from a servo control apparatus, state information including at least position deviation and a set of coefficients to be used by a backlash acceleration compensating part, by making the servo control apparatus execute a predetermined machining program, an action information output part for outputting action information including adjustment information on the set of coefficients included in the state information to the servo control apparatus, a reward output part for outputting a reward value in the reinforcement learning on the basis of the position deviation included in the state information, and a value function updating part for updating an action-value function on the basis of the reward value output by the reward output part, the state information and the action information.

A method for backlash compensation in motion control systems includes homing a payload of a motion control system, and performing a tooth-pitch non-uniformity procedure on the motion control system to identify a non-uniformity correction. A backlash lookup table is generated for use in backlash correction during normal operation of the motion control system. The backlash lookup table is generated using a training process that includes selecting a move sequence for operating the motion control system, and executing the move sequence with the non-uniformity correction. The training process further includes computing a backlash measurement describing backlash of one or more components of the motion control system during the move sequence, and storing the backlash measurement in the backlash lookup table.

To perform reinforcement learning enabling to prevent complicated adjustment of coefficients of backlash compensation and backlash acceleration compensation. A machine learning apparatus includes a state information acquiring part for acquiring, from a servo control apparatus, state information including at least position deviation and a set of coefficients to be used by a backlash acceleration compensating part, by making the servo control apparatus execute a predetermined machining program, an action information output part for outputting action information including adjustment information on the set of coefficients included in the state information to the servo control apparatus, a reward output part for outputting a reward value in the reinforcement learning on the basis of the position deviation included in the state information, and a value function updating part for updating an action-value function on the basis of the reward value output by the reward output part, the state information and the action information.

Stop and start detection block (63) determines whether or not a joint portion is in a stopped state before a rotation direction of the joint portion is inverted based on stop flag signal (Stop_Flg). When it is determined that the joint portion is in the stopped state, filter processing block (65) changes a frequency component of a correction amount for correcting backlash to a low frequency lower than a predetermined threshold value.