Although 3D printing technology has developed
rapidly in recent years, there are many problems to be solved.
The accuracy of the objects obtained by 3D printing is low
compared with subtractive manufacturing methods. There are
many factors that affect the errors, such as the shape of the
object, the properties of the material, and parameters in the
printing process. We takes Digital Light Processing (DLP) 3D
printing as an example in this paper and take the dental crowns
as the test category. We use the Convolutional Auto-Encoder
(CAE) architecture to make prediction and compensation for
the error of the 3D models directly. We use a simulation method
to obtain the data, and test on nonlinear deformation. The
results show that our network has a good ability to approximate
nonlinear deformation.