RBFNN Based Linear Motor Cogging Force Identification for Lithography Machines

摘要

The cogging force of linear motors significantly affects the positioning accuracy in ultraprecision lithography machines. Optimal motor design cannot completely eliminate it, leaving the realtime active compensation as the only means to counteract its adverse effect. This requires the real-time identification of the cogging force, which is however a challenging issue. Traditional approaches are either computationally too expensive or practically infeasible. This paper proposes to develop a two-stage trained RBF neural model for the detent force identification of the ultra-precision wafer stage platform in a lithography machine. Experimental results confirm the effectiveness of this data-driven method.