An intelligent self-learning method for dimensional error pre-compensation in CNC grinding

The problem of error compensation in manufacturing industry becomes important since the product quality is the manufactures' competition. This paper presents an intelligent self-learning method for dimensional error pre-compensation in CNC grinding. Measurements of the system output obtained from the previous runs are used to compute the correction for the current run without in-process sensing and measurement, which is attractive for applications lacking in situ measurements. A modified NC program is fed to the machine tool to push the system output towards target. The simulation results and the real system input and output responses show the feasibility and effectiveness of this intelligent self-learning method.