Predictive Control for Varying Sampling Period System in NCS

A predictive control based on online LSSVM (least squares support vector machine) is proposed to aim at the problem of varying sampling period in NCS (networked control system). Utilizing training data window and training data threshold, an online training method which adapts to control systems is developed based on LSSVM. When kernel function is linear in online LSSVM, the solution of control variable can be solved integrating the method of predictive control. The simulations show that the method has excellent robustness to varying sampling period with time-delay and packer loss in NCS.