针对考虑舵机特性的船舶航向离散非线性控制系统,提出了一种基于神经网络的自适应控制设计方法。为了消除离散系统后推设计中存在"因果矛盾"的问题,原船舶航向离散系统通过变换得到等价的能够预测变量的前向预测系统。通过使用单一神经网络逼近系统的所有未知部分,该控制设计方法可以有效地减轻控制系统存在的"计算量膨胀"问题,并具有控制器结构简单,控制参数少,易于工程实现等优点。同时,稳定性分析证明闭环系统的所有信号一致最终有界,并能使得航向跟踪误差任意小。最后,运用"育鲲"轮进行仿真研究以证明所提方法的有效性。
In this study, a novel adaptive control method was investigated for a ship course discrete-time nonlinear system with rudder dynamics based on a neural network. To solve the problem of discrete-time backstepping design procedures, the ship course discrete-time system was transformed into an equivalent ahead predict system in which the state variables could be predicted. The proposed control design scheme was shown to effectively solve the "explosion of complexity" problem by approximating all the unknown parts of the system using a single neural network. By this way, the control parameters are fewer, the controller structure is much simpler and the controller is easier to implemented in application. All the signals in the closed-loop system were shown to be uniformly ultimately bounded, ensuring that the tracking error converged to a small neighborhood at the origin. Finally, simulations of the vessel "Yukun" were used to demonstrate the effectiveness of the proposed algorithm.