摘要
采用一种Laguerre网络与SBP神经网络构成的组合模型和即时线性化方法实现了对Wiener非线性系统的自适应预测控制策略 .组合模型无需动态系统的阶次和时延的结构先验知识 ;即时线性化 ,即是在线依据每个控制周期由模型获得的系统未来各步预测输出 ^y0 (k+i) k,(i=1 ,… ,P)对非线性模型进行线性化 ,进而利用线性化模型进行控制优化求解 即时线性化避免了非线性控制优化求解的困难 .文中还给出了实现即时线性化的算法 。
A new predictive control scheme for Wiener type non li near systems by the combination Laguerre networks with SBP neural networks (SBP NN) and the method of immediate linearization (IL) is presented. The combined mo del does not need the order and the delay of the dynamic system; the IL, which i s to performt the linearization of the non-linear model at every control period by the future predictive output obtained from the model at each step online, can avoid the difficulty of solving non-linear optimization. Simulation results show that the algorithm and the scheme are effective here.
基金
安徽省自然基金资助项目 (No .990 43 2 0 3 )
