In this article, we investigate the horizontal trajectory tracking problem for an underactuated stratospheric airship subject to nonvanishing external disturbances and model uncertainties. By transforming the tracking errors into new virtual error variables, we can specify the transient and steady-state tracking performance of the resulting nonlinear system quantitatively, which means that under the proposed control scheme, the tracking errors will converge to prescribed residual sets around the origin before a preselected finite time with decay rates no less than a preassignable value. To address unknown items, minimal learning parameter (MLP) techniques for neural networks (NNs) approximation are employed, which efficaciously relax the computational burden, enhance the robustness against dynamics uncertainties and provide an improved property for disturbances rejection. A finite-time convergent observer (FTCO) is incorporated into the control framework to realise output-feedback control, ensuring that estimation errors are bounded during operation and approach zero within a finite time. Stability analysis proves that all the closed-loop signals are uniformly bounded. The effectiveness and advantages of the proposed control strategy are verified by simulation results.

中文翻译：

---

具有规定性能的平流层飞艇的神经自适应输出反馈轨迹跟踪控制

在本文中，我们研究了受非消失外部干扰和模型不确定性影响的欠驱动平流层飞艇的水平轨迹跟踪问题。通过将跟踪误差转化为新的虚拟误差变量，我们可以定量地指定所得非线性系统的瞬态和稳态跟踪性能，这意味着在所提出的控制方案下，跟踪误差将收敛到原点周围的规定残差集在衰减率不小于可预分配值的预选有限时间之前。为了解决未知项目，采用了神经网络 (NNs) 逼近的最小学习参数 (MLP) 技术，有效地减轻了计算负担，增强对动态不确定性的鲁棒性，并为抗扰提供改进的特性。在控制框架中加入有限时间收敛观测器（FTCO），实现输出反馈控制，确保估计误差在运行过程中是有界的，并在有限时间内趋近于零。稳定性分析证明所有闭环信号都是一致有界的。仿真结果验证了所提控制策略的有效性和优势。