Observer-Based Adaptive NN Tracking Control for Nonstrict-Feedback Systems with Input Saturation

Input saturation is one of the common phenomena in many practical systems, and it is main obstacles that limits the systems performance. In this paper, the adaptive neural network (NN) control problem has been discussed for a family of uncertain nonstrict-feedback systems with input saturation. The innovations are summarized as follows: (1) the auxiliary systems and the NN state observer are developed to eliminate the influence of input saturation and estimate unmeasurable states; (2) in order to against the drawback of “explosion of complexity" for the traditional backstepping control technique (BCT), the dynamic surface control technique is used to reduce the excessive computation burden; (3) the proposed NN control approach for nonstrict-feedback systems only utilize the property of radial basis function-neural networks (RBF-NNs), instead of the restrictive assumption. Furthermore, unknown smooth functions are approximated by RBF-NNs in nonlinear systems. By employing the BCT, an adaptive output-feedback controller has been constructed. Meanwhile, all signals in closed-loop system are semi-globally uniformly ultimate bounded. An explicit function with the saturation error and designed parameters is obtained, which indicates the tracking error can be tuned through the saturation error and designed parameters. Finally, the superiority of the proposed control technique is validated by two examples.