DECENTRALIZED ADAPTIVE CONTROL FOR A CLASS OF NONLINEAR CONTINUOUS TIME INTERCONNECTED SYSTEMS USING NEURAL NETWORKS

The authors present a completely decentralized adaptive control for a class of nonlinear, continuous time, interconnected systems, in the sense that no information exchange between the subsystems is needed. The main idea consists of predicting the interconnection output terms using the polynomial series. These predictions are then used in the local control law, which offers a general solution in the area of interconnected linear and nonlinear systems with arbitrary interconnections. In the present work, we consider the case of controlling a class of interconnected systems using neural networks. A multilayer neural network is used to model each unknown subsystem and generate the control law. Based on the error between the plant output and the model output, the weights of the neural network are updated online according to a gradient learning rule with dead zone. Finally, the results are illustrated by numerical examples.