基于SVM阶逆系统方法的非线性预测控制

Nonlinear Predictive Control Based on SVM αth-Order Inverse Method

摘要:

针对一类未知非线性大时延系统,利用SVM特有的表达任意非线性映射能力,辨识得到非线性系统的α阶逆模型,然后将其串在原系统前组成SVMα阶伪线性复合系统。以此复合系统为被控对象,用预测控制方法实现对其有效控制。该方法不仅将非线性系统简化成伪线性系统,而且具有良好的控制效果,控制器设计简单。

Abstract:

For the control of a class of model unknown large time-delay nonlinear systems, a novel predictive controller based on support vector machine which could express any nonlinear mapping was proposed. This method used SVM identify the nonlinear system’s αth-order inverse model and combine it in front of the former system to form a pseudo-linear composite system with minimum time-delay. Then, a linear predictive controller was designed for the pseudo-linear composite system. Since the complicated nonlinear controlled system was simplified into a linear composite system, the predictive controller was easy to be implemented. And it was easy and simple to analyze the stability and stable performance of the predictive control system.