Method and apparatus for correcting positioning errors on a machine tool

A machine tool having a plurality of components subject to thermal deformation, for machining a workpiece using a tool member in which said tool member and workpiece are moveable relative to each other, said machine tool comprising:

• a numerical controller for providing a drive signal to move said tool member relative to said workpiece along at least one axis of movement;
• thermosensors provided at the different positions of said machine tool for sensing a plurality of temperatures at said positions of said machine tool;
• an inferring means for inferring a positioning error of said tool member relative to said workpiece in the direction of said axis of movement, said inferring means including a neural network having a plurality of neurons, said neural network being operable to receive temperature signals indicative of the sensed temperatures and to thereupon generate a signal indicative of a positioning error in the direction of said axis of movement;
• a correction value calculator, responsive to said positioning error signal, for providing a correction signal indicative of a correction value in the direction of said axis of movement;
• and
• wherein said numerical controller is operable to correct said drive signal using said correction signal to correct for positioning errors due to thermal deformation in the direction of said axis of movement.

A method of correcting for positioning errors between a workpiece and a tool member in a machine tool having a plurality of components subject to thermal deformation, wherein the tool member and workpiece are adapted to move relative to each other along at least one axis of movement under control of a drive signal from a numerical controller, said method comprising:

• sensing the temperatures at a plurality of positions of said machine tool;
• inputting the sensed temperatures to at least one neural network having a plurality of neurons;
• inputting signals indicative of thermal deformation along said axis of movement associated with said sensed temperatures to said neural network to thereby preset bonding strengths between said neurons associating said sensed temperatures and said thermal deformations;
• thereafter, inferring thermal deformations along said at least one axis based on said sensed temperatures and on said preset bonding strengths to thereby generate a thermal deformation correction value;
• and
• correcting positioning errors responsive to said thermal deformation correction value.