Backpropagated adaptive critic neurofuzzy controller for nonlinear dynamic system

In this paper, a new paradigm of a neurofuzzy controller is developed for learning to refine a rule-based fuzzy logic controller. Two architectures are used: a one step neural network predictor and the backpropagated adaptive critic neurofuzzy controller. The backpropagated adaptive critic neurofuzzy controller algorithm is developed and used to train the neurofuzzy controller. The control mechanism is then analysed and simulation results show that the scheme has a high capability to learn and generalize, and can deal with large unknown nonlinearities