Free-model based adaptive inverse neurocontroller for dynamic systems

The newly developed “free-model” concept is applied to the adaptive inverse controller and identifier, and implemented in neural networks to make nonlinear dynamic systems to track desired outputs. The free model based neural network converges faster than the conventional neural network representation. Stability conditions are developed for training. Using stability conditions and the free model based neural network, a model reference adaptive inverse control scheme is developed for nonlinear plants with satisfactory performance