Critic-Only Learning Based Tracking Control for Uncertain Nonlinear Systems with Prescribed Performance

A critic-only learning-based tracking control with prescribed performance was proposed for a class of uncertain nonlinear systems. Based on an estimator and optimal controller, novel controller designed to make errors uniformly ultimately bounded limited in region. First, unknown system dynamic employed online approximate the uncertainty invariant manifold. Subsequently, by running cost function, derived learning neural network, which ensured that can evolve within area while minimizing function. Specifically, weight update be driven estimation error, avoiding introducing actor-critic architecture complicated law. At last, stability closed-loop analyzed Lyapunov theorem, evolved controller. The effectiveness demonstrated two examples.