A switch is turned on by a learning control start command from a master control
unit, and the positional deviation at respective cycles is read in. Correction
data read out from a learning memory is added to the positional deviation, and
the result is filtered by band limiting filter and then stored in the learning
memory as correction data. The correction data read out from the memory is compensated
for phase delay, fall in gain, and the like, by a dynamics compensating element,
and is added to the positional deviation and input to a positional control section.
When the command pattern for the same shape is completed, and a learning control
end command is output, whereupon the switch is turned off and learning control terminates.
