GMDH-based autonomous modeling and compensation for nonlinear friction

This paper presents a novel mathematical model-based compensation algorithm for the nonlinear friction in table drive systems using the group method of data handling (GMDH). In the proposed scheme, the nonlinear friction can be autonomously modeled as a polynomial expression for appropriate control state variables according to the process of GMDH and, as a result, the complicated structural modeling and its parametrization, indispensable in conventional model-based strategies, can be completely eliminated. In addition, the proposed GMDH-based model can achieve the generalization ability for table drive condition and, as a result, the robust compensation for friction can be attained against the change of drive conditions. Experimental results using a table drive system of actual machine tools show the significant performance improvement of the proposed algorithm in the trajectory control with velocity reversal motion.