The Jiangsu Provincial Key Laboratory of Networked Collective Intelligence, and School of Mathematics, Southeast University, Nanjing 211189, China.
Neural Netw. 2021 Oct;142:288-302. doi: 10.1016/j.neunet.2021.05.014. Epub 2021 May 18.
The event-triggered adaptive neural networks control is investigated in this paper for a class of fractional-order systems (FOSs) with unmodeled dynamics and input saturation. Firstly, in order to obtain an auxiliary signal and then avoid the state variables of unmodeled dynamics directly appearing in the designed controller, the notion of exponential input-to-state practical stability (ISpS) and some related lemmas for integer-order systems are extended to the ones for FOSs. Then, based on the traditional event-triggered mechanism, we propose a novel adaptive event-triggered mechanism (AETM) in this paper, in which the threshold parameters can be adjusted dynamically according to the tracking performance. Besides, different from the previous works where the derivative of hyperbolic tangent function tanh(⋅) needs to have positive lower bound, a new type of auxiliary signal is introduced in this paper to handle the effect of input saturation and thus this limitation is released. Finally, two numerical examples and some comparisons are provided to illustrate our proposed controllers.
本文针对一类具有未建模动态和输入饱和的分数阶系统（FOSs），研究了事件触发自适应神经网络控制。首先，为了获得辅助信号，并避免未建模动态的状态变量直接出现在设计的控制器中，将指数输入状态实用稳定性（ISpS）的概念和一些相关引理从整数阶系统扩展到 FOSs。然后，基于传统的事件触发机制，本文提出了一种新的自适应事件触发机制（AETM），其中阈值参数可以根据跟踪性能动态调整。此外，与以前的工作不同，以前的工作中需要正的下界来保证双曲正切函数 tanh(⋅)的导数，本文引入了一种新型辅助信号来处理输入饱和的影响，从而释放了这一限制。最后，提供了两个数值示例和一些比较，以说明我们提出的控制器。
