Robust control based on LS-SVM for uncertain nonlinear system

In this paper, a new stable robust adaptive control approach is presented for SISO uncertain nonlinear system. The key assumption is that LS-SVM approximation errors and external disturbances satisfy certain bounding conditions. The LS-SVM can find a global minimum and avoid local minimum. Its weights in the observer can be tuned after training phase and find optimistic value automatically. By combining LS-SVM, the system state vector is estimated by an observer efficiently. A simulation example demonstrates the feasibility of the proposed approach.