This article investigates the finite-time control of the strict-feedback nonlinear system using composite learning based on the historical stack. The controller design adopts the backstepping scheme while the nonlinear function is introduced to avoid the singularity problem. The first-order Levant differentiator is introduced to obtain the filtered command signal and the compensation signal is further constructed. To indicate the learning performance, the historical data over the moving time window are analyzed to construct the predictor error using the maximum–minimum singular value algorithm. Furthermore, the finite-time neural update law is proposed. The stability of the closed-loop system is analyzed via the Lyapunov approach. The performance of the proposed method is verified using simulations.

中文翻译：

---

使用历史堆栈的严格反馈非线性系统的有限时间复合学习控制


本文研究了使用基于历史堆栈的复合学习对严格反馈非线性系统的有限时间控制。控制器设计采用反步方案，同时引入非线性函数以避免奇异性问题。引入一阶Levant微分器来获得滤波后的命令信号，并进一步构造补偿信号。为了指示学习性能，分析移动时间窗口上的历史数据，以使用最大最小奇异值算法构建预测误差。此外，提出了有限时间神经更新律。通过Lyapunov方法分析闭环系统的稳定性。使用模拟验证了所提出方法的性能。