Selection and modeling of temperature variables for the thermal error compensation in servo system

In order to compensate the thermal error of the coordinate measuring machine accurately, selecting the thermal key points for building the error model is important. A method based on the multivariable statistics analysis was described in the thermal error compensation process of servo system. The measurement points are selected based on the cluster analysis theory and the optimal regression model which are given by stepwise regression. The compensation model is made for servo system. We use PT100 platinum resistance to measuring the thermal change of the selected points for building the model. The result shows that, the methods not only avoid the correlation of the measurement points and ensure the accuracy of the model, but also save the time and cost. According to this conclusion, it is useful to decreasing and restraining the thermal errors, and to present a clearer and deeper knowledge of selection strategy of the thermal points.