科研之友
微信
新浪微博
Facebook
分享链接
摘要
Both of deterministic and probabilistic sparse kernel learning (SKL) methods are effective for process modeling. Based on the SKL model with a polynomial kernel, a simple nonlinear control strategy is investigated. First, an SKL identification model is obtained using a polynomial kernel. Then, a predictive control performance index, which is characterized as an even-degree polynomial function of the manipulated input, is formulated. Consequently, the optimal manipulated input can be efficiently obtained by solving a simple root problem of an odd-degree polynomial equation because of its special structure. A comparative study on a benchmark problem shows its superiority to traditional controllers. Also, some attributes of the proposed control strategy can result in a practicable solution for real-time control.
