摘要
本文提出了二种基于对象CARMA和CARIMA模型的自适应广义预测控制器。参数估计采用变遗忘因子的最小二乘法,用加权预测控制律代替一步预测控制律,充分利用预测控制信号,减少了错误控制信号作用,从而改进了现有的广义预测控制算法.仿真表明该控制器适用于非最小相位系统和开环不稳定系统,对系统阶次和时延有很好的鲁棒性,有较强的抑制干扰和容错控制能力,改善了控制品质。
In this paper, two adaptive generalized predictive controllers based on CARMA and CARIMA models are proposed. The new algorithms reduce the effect of control input errors and the possibility of input ocsillation and indeed saturation by taking into account a number of predictions of the input signal at each time instant, the actual signal applied being a weighted summation of all those available and by using a least-squares estimator with a variable forgetting factor for adaptive implementation. Simulation shows that the controllers have a strong robustness to the change of dead-time, model order and disturbance, they are suitable for systems that are both unstable and/or in non-minimum phase and for fault-tolerent control, they improve properties of the controlled system.
出处
《控制与决策》
EI
CSCD
北大核心
1991年第1期7-13,共7页
Control and Decision
