IEEE Trans Neural Netw Learn Syst. 2021 Dec;32(12):5416-5426. doi: 10.1109/TNNLS.2020.3027689. Epub 2021 Nov 30.
This article investigates an adaptive finite-time neural control for a class of strict feedback nonlinear systems with multiple objective constraints. In order to solve the main challenges brought by the state constraints and the emergence of finite-time stability, a new barrier Lyapunov function is proposed for the first time, not only can it solve multiobjective constraints effectively but also ensure that all states are always within the constraint intervals. Second, by combining the command filter method and backstepping control, the adaptive controller is designed. What is more, the proposed controller has the ability to avoid the "singularity" problem. The compensation mechanism is introduced to neutralize the error appearing in the filtering process. Furthermore, the neural network is used to approximate the unknown function in the design process. It is shown that the proposed finite-time neural adaptive control scheme achieves a good tracking effect. And each objective function does not violate the constraint bound. Finally, a simulation example of electromechanical dynamic system is given to prove the effectiveness of the proposed finite-time control strategy.
本文研究了一类具有多个目标约束的严格反馈非线性系统的自适应有限时间神经控制。为了解决状态约束和有限时间稳定性出现带来的主要挑战，首次提出了一种新的障碍李雅普诺夫函数，不仅可以有效地解决多目标约束问题，而且可以确保所有状态始终在约束区间内。其次，通过结合命令滤波方法和反推控制，设计了自适应控制器。更重要的是，所提出的控制器具有避免“奇点”问题的能力。引入补偿机制来中和滤波过程中出现的误差。此外，在设计过程中使用神经网络来逼近未知函数。结果表明，所提出的有限时间神经自适应控制方案实现了良好的跟踪效果，并且每个目标函数都不违反约束边界。最后，给出了机电动力学系统的仿真示例，以证明所提出的有限时间控制策略的有效性。
