Adaptive compensation of friction error for numerical control machine tool

To improve the machining precision of numerical control machine tool and reduce contour error caused by friction, a novel adaptive compensation of friction error strategy is proposed to solve the problem that conventional method for compensation of friction error can hardly achieve desired performance. Via coarse and fine learning process for amplitude of compensation pulse, the amplitude function of friction compensation pulse is established by precision curve fitting based on generalized regression neural network algorithm, and the friction error can be compensated adaptively. The new compensation strategy is adopted on a X-Y worktable, the peak of circular contour error in quadrant and the evaluation function value of friction compensation are reduced by more than 55% under different working conditions. Compared with the conventional rectangular pulse friction compensation, the proposed approach effectively improves the machining precision of the machine tool due to the robust adaptability, lower impact on the servo feed system and better compensation performance.