Design of Indirect Adaptive Neural Control Systems
Ieroham Baruch, Joså Martin Flores*, Boyka Nenkova
Institute of Information Technologies, 1113 Sofia
*Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional, 07360 Mexico D.F., MEXICO
Abstract: A parametric Recurrent Neural Network (RNN) model and an improved dynamic Back-propagation (BP) method of its learning are applied for nonlinear plants identification and state estimation. The obtained parameters of the RNN model are used for an adaptive control system design. The paper suggests three main types of state-space control with RNN states estimation: a proportional (P); a proportional plus integral (PI) and an trajectory tracking control. The applicability of the proposed neural adaptive control schemes is confirmed by simulation results.
Keywords: recurrent neural networks, systems identification, backpropagation-through-time learning, state estimation, adaptive control, nonlinear systems.