Study of the thermal deformations on a coordinate measuring machine and compensation via a neural network model

摘要

Coordinate measuring machines (CMMs) are a special kind of machine tool with their own specific thermal behavior. The difference is very important when analyzing the thermal behavior of a CMM. The investigations on the relationship between temperature variation and thermal deformation in CMMs show that the ambient temperature is a major influencing factor. Thermal errors appear at environment temperatures other than 20℃. This paper illustrates that thermal errors and nonlinear displacements on the CMM are related to the machine structures. The effects of internal heat sources and temperature gradients are also described. The applicability of a neural network model to compensate for thermal deformations is examined. The paper presents the different phases of data measuring, learning, training, and thermal error prediction with a neural net algorithm. With the use of additional test data files, the prediction power of the calculated model is shown. A general compensation diagram is added that summarizes the data flow when using the presented ideas.