基于改进LS-SVM的网络化预测控制

    Networked Predictive Control Based on Improved LS-SVM

    • 摘要: 针对存在网络延时的网络化控制系统, 提出一种具有最大预测步数的网络化预测控制策略.首先, 分析网络延时的组成, 确定网络延时的最小值, 并利用自由权矩阵方法计算网络延时最大允许上界, 作为最大预测步数;进而, 利用改进的最小二乘支持向量机在线预测网络延时;最后, 利用基于往返延时的补偿策略选择合适的预测控制量来补偿网络延时的影响.仿真结果表明该控制策略有效.

       

      Abstract: A networked predictive control strategy with maximum predictive horizon was proposed to improve the performance of networked control systems with network delay in this paper. First, the composition of network delay was analyzed to determine the minimum delay, and the maximum allowable delay bound, which was used as the maximum predictive horizon, was determined by the free weighting matrix method. Second, an improved least squares support vector machine was used to predict the network delay online. Finally, the appropriate predictive control vector was chosen by the internet delay prediction method to compensate for the networked delay. Simulation results show that the proposed method is effective.

       

    /

    返回文章
    返回