不确定时滞系统的自适应支持向量机Smith预估控制

SVM-Based Adaptive Smith Predictive Control of Uncertain System with Time Delay

针对不确定时滞系统提出了一种自适应支持向量机Smith预估控制新方法.首先采用支持向量机对被控对象进行建模,然后设计了一个自适应支持向量机Smith智能预估器,解决了传统Smith预估控制需要预先知道被控对象精确数学模型的问题,克服了基于神经网络的Smith预估控制的不足.仿真实验结果表明,自适应支持向量机Smith预估控制方法充分利用了支持向量机的非线性映射能力,在被控对象数学模型未知的情况下对不确定时滞对象进行控制,具有良好的控制品质,特别是当对象特性发生变化时,还具有良好的适应性.

<Abstrcat>A new SVM-based adaptive Smith predictive control method for uncertain system with time delay was proposed. Firstly, the plant was modeled with SVM. Then, an SVM-based adaptive Smith intelligent predictor was designed. This new approach, unlike the conventional Smith predictor which needs the precise mathematical model of the plant, overcame the drawback of Smith predictive control based on neural networks. Simulation results showed that this new method had a good control performance by making full use of the nonlinear map capability of SVM when the plant model was unknown, and especially, when the plant was time varying. And this approach was ideally adaptable.