Neural-Networks-Based Prescribed Tracking for Nonaffine Switched Nonlinear Time-Delay Systems
针对一类未知的非仿射切换非线性时滞系统，利用神经网络分离和逼近技术，提出一种自适应方案实现预设跟踪性能，无需每个子系统的可控性条件。
In this article, by using the neural-networks (NNs) separation and approximation technique, an adaptive scheme is presented to deliver the prescribed tracking performance for a class of unknown nonaffine switched nonlinear time-delay systems. The nonaffine terms are indifferentiable and the controllability condition is not required for each subsystem, which allows the considered tracking problem to not be efficiently solved by the traditional adaptive control algorithms. To solve the problem, NNs are utilized to separate and approximate the nonaffine functions, and then the dynamic surface control and convex combination method are utilized to construct a controller and a switching strategy. In addition, an adaptive law is considered for each subsystem to reduce the conservativeness. Under the designed controller and switching strategy, all the signals of the resulting closed-loop system are bounded, and the tracking performance is achieved with a prescribed level.