Prediction of positional errors of a three axis machine tool using a neural network

The accuracy of a part being machined depends directly on the total error in position of the cutting tool with respect to the workpiece. This error is due to a combination of geometric and thermally induced errors that occur in the structural elements of the machine tool. In this paper, the positional error map of a three axis machine tool is predicted using a neural network based on the back-propagation algorithm. The inputs to the network are the three coordinates of the cutting tool and a bias input corresponding to each of these coordinates which is used to account for the direction of motion of each slide. The outputs from the network are the components at the tool point positioning error at this point in the workspace. These bias nodes account for the reversal errors that occur in machine tools. Three different training sets were created using the kinematic model of the machine tool with different methods in sampling the data. Computer simulations show that the neural net is able to learn the error map of the three axis machine tool accurately with excellent generalization properties in both ideal and noisy environments.