Abstract

This thesis presents a constrained optimization-based neuro-adaptive controller (CoNAC) for uncertain Euler-Lagrange systems subject to weight norm and input constraints. A deep neural network (DNN) is employed to approximate an ideal stabilizing control law which compensates for lumped system uncertainties while addressing both types of constraints. The weight adaptation laws are derived from constrained optimization theory, ensuring first-order optimality conditions at steady state. The controller’s stability is rigorously analyzed using Lyapunov theory, ensuring bounded tracking errors and DNN weights. Two numerical simulations were constructed to compare CoNAC with other benchmark controllers. The simulations demonstrated effectiveness of CoNAC in tracking error regulation and satisfaction of constraints.