PURPOSE: To provide the learning controller which performs stable feedforward control according to variation of a controlled system which vary in charactertistics affecting the control in a short time and includes nonlinearity etc.

CONSTITUTION: A command for the output of the controlled system 2 is inputted to a module 5 to perform operation for approximating the controlled system 2 to the command; and the current and past outputs and inputs of the controlled system 2 or a signal indicating the characteristic variation of the controlled system 2 is supplied to a module 6 and the output of the module 5 is controlled corresponding to the output of the module 6 and outputted as part of the input to the controlled system 2, thereby performing feedback control by a feedback controller 1.