Decentralized Saturation-Tolerant Control for Stochastic Interconnected Systems with Preassigned Performance and Input Quantization

This paper proposes a decentralized saturation-tolerant control solution for interconnected stochastic systems subject to input quantization. Firstly, a new saturation-tolerant prescribed performance control approach, with input-relevant terms and time-varying functions, is presented to accomplish preassigned performance and prevent the violations of prescribed behaviors in face of input saturation, where the input-relevant terms generated by auxiliary systems are to provide timely compensations on control constraints. Secondly, a signal-parameter-updating neural network (SNN) is derived such that the issue of parameter updating explosion induced by high learning dimension is effectively handled. Then aiming at lessening the transmission cost in control channel, a hysteresis quantizer (HQ) is embedded in our controller design, where the amount of data transmitted is greatly reduced while control precision cannot be evidently weakened due to constraints on tracking errors by performance boundaries. Finally, simulations with comparisons are performed that validate the effectiveness of proposed methods.