Novel neural adaptive terminal sliding mode control for TCP network systems with arbitrary convergence time

作者：Qi, Xuelei; Li, Chen; Chen, Bao; Ni, Wei; Ma, Hongjun^*

来源：NEURAL COMPUTING & APPLICATIONS, 2023, 35(27): 20365-20374.

DOI：10.1007/s00521-023-08746-z

• Network congestion control
• Terminal sliding mode control
• Arbitrary convergence time
• Neural network

The problem of adaptive arbitrary convergence time terminal sliding mode control (TSMC) for nonlinear networked systems with completely unknown functions and nonresponsive UDP stream disturbances is studied. In this work, an improved sliding surface based on arbitrary convergence time is designed to overcome the issue that the convergence rate depends on the system state. Different from the traditional TSMC, this paper introduces holistically an auxiliary term (nonautonomous differential equation) into the design of a sliding mode controller, which can compensate for the influence of neural adaptive approximation error and ensure the convergence of a closed-loop system at prescribed time. A robust compensation signal is also constructed by introducing a super twisting disturbance observer to process nonresponsive UDP flows. In addition, the controller is designed in a piecewise continuous manner, so that the sliding mode surface can be selected arbitrarily, according to the adaptation after the prescribed time. The effectiveness of the new controller is verified by extensive simulations in comparison with related works.

• 单位
  测试单位; 东北大学