Thermal errors of the machine tools have a
significant effect on the machining precision. In this paper,
temperature variables selection based on the fuzzy c-means
cluster analysis is studied, robust regression theory is utilized
to establish the relationship between the thermal errors and
the temperature variables, and large residuals are given small
weights and leave the residuals associated with extreme points.
Pt thermal resistances are used to measure the temperature
change and the eddy current sensors are used to monitor the
thermal shifts of the spindle, the test results show that robust
regression method can predict the thermal errors of the
machine accurately. The coupling among the variables is also
solved, which can be used for the error compensation of the
machine tool so as to meet the accuracy demands of the
precision machining.