Friction Modeling And Compensation By SVM

Friction caused low speed creep and tracking errors which presents in servo systems has become an impediment to improve their performance .In order to eliminate or reduce the effect of friction, and increase the system performance, appropriate methods must be carried out. This dissertation focuses on the following work:Firstly, the present situation of friction modeling and compensation at home and abroad are summarized. Friction modeling and compensation using support vector machine(SVM) is put forward.Secondly, a improvd SVM-- the least squares support vector machine(LS-SVM) is introduced, and friction modeling and compensation by LS-SVM in servo systems.Finally, a mathematic model for a typical mechanical position servo system with friction model by LS-SVM is established. A backstepping control law by LS-SVM is derived with Lyapunov stability theory to ensure astringency of parameter estimation and global stability of servo systems with friction. Simulation results show that system robustness and tacking accuracy are improved with proposed method than that with traditional PID control.