Neural network control using composite learning for USVs with output error constraints

In this paper, by focusing on trajectory tracking control of unmanned surface vessel (USV), we present a control method considering uncertain dynamics and output error constrains. Firstly, by using the properties of tan-type barrier Lyapunov function (BLF), the output tracking error can be constrained. Secondly, we use radical basis function neural network (RBF NN) to approximate the uncertain dynamics. Considering that the estimated parameters convergence cannot be achieved in the absence of persistent excitation (PE) conditions, the composite learning update law of the weight matrix in the NN is adopted to guarantee the parameters convergence under interval excitation (IE) conditions which is easier to reach. In simulation studies, it is proven that the USV have good ability to follow the pre-designed trajectory with small tracking error and the parameters convergence can be ensured.