New control strategy based on identification and prediction using neural network
Neural Adaptive Control. Neural Networks Based Control. Neural Networks. Neural Networks Based Identification of Dynamic Systems.
Neural networks have been shown to be a tool very useful in the adaptive control area, contributing to its growth. Several authors have proposed technics based on neural networks that were to supplement or replace a previous one in order to cover more classes of systems that can be controlled. This work has the main contribution the proposal of a control law which generalizes an existing one in a hybrid neural adaptive control scheme. An analysis of the old control law was made and some comparative tests were performed showing that the generalization was achieved. The tests were performed using an MLP neural network as an identifier by the NNARX model structure. Although we intend to conduct further studies and some ideas to try to improve the proposed control law are yet to be realized.