Adaptive neural control for coordinated rolling motion of vessels with zero speed via composite learning

Abstract

An intelligent control method under the backstepping and dynamic surface control (DSC) design framework is proposed in this article. The proposed method is aiming at solving the problem of coordinated rolling motion control of the supply and receiving vessels with internal dynamic uncertainty, unknown external nonlinear disturbance, input saturation, and unmeasurable rolling angular velocity. In the control design, a neural-based state observer is constructed to generate the unmeasurable rolling angular velocity, which achieves a separate design from the control law. To compensate the input saturation effect, a second-order auxiliary dynamic system (ADS) is designed, and its states are inset into the transformation of coordinates required by backstepping. To realize the accurate reconstruction of the neural network for the dynamic uncertainty, the adaptive neural and nonlinear disturbance observer (NDO) are used to reconstruct the internal dynamic uncertainty and unknown external disturbance, respectively. And then the serial–parallel estimation model is introduced to establish a predictor of the system state, and a composite adaptive learning law is designed, which includes the velocity error and predictor error. As a result, the idea of classification reconstruction is developed. Finally, an adaptive neural composite learning coordinated rolling motion control solution is developed. Lyapunov stability analysis shows that all signals in the closed-loop control system are bounded. The effectiveness of the control solution is verified by the simulations.