基于结构改进Smith预估单神经元PSD的长时延NCS研究

针对长时延网络控制系统（NCS）中存在时延和被控对象等问题,提出了一种新型结构改进Smith预估单神经元PSD控制器.新型控制器由结构改进Smith预估器和单神经元PSD控制器组成,结构改进Smith预估器通过对传统Smith预估器模型结构上的改进,避免了建立网络时延的预估模型,无需对网络时延进行在线测量、估计或辨识,补偿了网络时延对系统的性能带来的影响;单神经元PSD控制器解决了结构改进Smith预估器的模型失配问题,通过Matlab环境下的TrueTime工具箱对NCS进行仿真实验,并对传统PID控制方法、结构改进Smith预估PID控制方法和结构改进Smith预估单神经元PSD控制方法进行了仿真比较.研究结果表明,该新型控制器有效地补偿了网络时延对NCS的性能带来的影响,在模型失配情况下具有良好的鲁棒性和适应性,是对长时延NCS-种可行且有效的控制方法.

NCS with long time-delay based on structure improved Smith predictor with single neuron PSD control

Aiming at the problems of network time-delay and controlled plant in network control system （NCS）with long time-delay,a new structure improved Smith predictor with single neuron PSD control was proposed.This new controller was composed of structure improved Smith predictor and single neuron PSD controller.Improved by classic Smith predictor model structural adjustment,the structure improved Smith predictor can be used for compensating the influence of network time-delay on the performance of NCS without establishing the prediction model of network time-delay,and network time-delay doesn＇t need to be measured,estimated or identified on line.The single neuron PSD controller can be used for solving the problems of structure improved Smith predictor model mismatch.The NCS simulation experiment with comparison between classic PID controller,structure improved Smith predictor with PID control and structure improved Smith predictor with single neuron PSD control was emulated by TrueTime toolbox in Matlab.The results indicate that this new controller effectively compensates the influence of network time-delay on the performance of NCS,and has good robustness and adaptability in case of model mismatch,so it is a feasible and effective control method for NCS with long time-delay.