Adaptive critic control and functional link neural networks

An adaptive critic neural control algorithm, employing an analytic relationship between the optimal cost function and the optimal control law is defined. This approach reduces the complexity of the adaptive critic algorithm, by replacing the iteration on the control law with the solution of a nonlinear equation, while simultaneously facilitating the development of stability and robustness criteria. The resultant algorithm can be implemented without a priori knowledge of the plant dynamics, yielding an adaptive control algorithm which does not require an explicit system identification process. Finally, the algorithm is specialized to the linear case where it is realized by a system of three coupled functional link neural networks.