Dual adaptive control of nonlinear stochastic systems based on echo state network

The dual adaptive control problem is addressed for a class of Single-In-Single-Out (SISO) stochastic, affine nonlinear, discrete systems. The nonlinear functions of system model are assumed to be unknown and approximated by the echo state networks (ESNs). The parameters of ESNs are online adjusted using the conventional Kalman filtering technique. The dual adaptive control law is designed considering an explicit-type, suboptimal cost function based on the innovations. The simulations testified the performance of the proposed control law.