基于在线LS-SVM的网络预测控制系统

Networked Predictive Control System Based on Online LS-SVM

  • 摘要: 针对网络控制系统(NCS)的鲁棒性,提出基于在线最小二乘支持向量机(LS-SVM)的预测控制方法.在LS-SVM的基础上,利用训练数据窗及训练数据阈值,推导出适合控制系统的在线训练方法.当在线LS-SVM的核函数取线性函数时,结合预测控制方法得到在线LS-SVM预测控制量的方程解,并将其应用于存在时延、丢包及包序错乱的NCS进行验证.仿真显示了该方法的快速性、准确性、鲁棒性.

     

    Abstract: For the robustness of networked control system(NCS),a predictive control method based on online least squares support vector machine(LS-SVM) is proposed.Based on LS-SVM,training data window and training data threshold are employed to develop an online training method suitable for the control system.When the kernel function is linear in the online LS-SVM,the control variable can be solved by integrating the method of predictive control.In order to make validation,the method is applied to the NCS which has time delay,data packet dropout and desequencing.The results of simulation show that the presented method is rapid,accurate and robust.

     

/

返回文章
返回