基于动态邻域微粒群的Smith预估双控制器设计

针对模型未知时滞系统的预测补偿控制,提出一种基于动态邻域拓扑微粒群算法以优化动态神经网络的参数,并将其作为预估器和辨识器应用于一种新的Smith预估双控制器结构设计.利用微粒群算法空间搜索能力指标,动态建立邻域拓扑结构,优化神经网络参数,并将两者的组合模型应用于新的双控制器结构,将负载扰动和定值控制分开,以提高Smith预测补偿模型的控制精度和鲁棒性,最后通过仿真验证了所提出方法的有效性.

Smith predictive double controllers design based on dynamic neighbor particle swarm optimization algorithm

For the unknown time-delay system of predictive compensation control,a dynamic neighborhood topology particle swarm optimization(PSO) algorithm is presented to optimize the parameters of dynamic neural networks,which is taken as a predictor and identifier in the new double-controller Smith predict structure,respectively.By using the particle swarm optimization space search capability index,the neighborhood topologies of PSO algorithm are dynamically created to optimize the neural network parameters.After that,the combination model is applied to the new two double-controller structure,which separates the load disturbance and fixed value control,and improves the control precision and robustness of Smith predictive compensation model.Finally,simulation results show the effectiveness of the proposed method.