NN-GA based printing parameters optimization for 3DP

Abstract

With the rapid printing speed and low cost, Three Dimensional Printing (3DP) is widely used. However, the dimensional accuracy of components are not perfect due to the shrinkage and deformation of component after the printing and post-processing. This study analyzes the factors affect the printing accuracy in 3DP and determines the range of shrinkage of the printing process. Neural Network (NN) is used to describe the complicated relationship between the dimensional accuracy of component and printing parameters. In order to minimizing dimensional error of specimen, Genetic Algorithm (GA) is used to optimize the 3DP print parameters such as binder saturation, the layer thickness and shrinkage compensation in X, Y and Z directions respectively. The four experiments with default parameters, the limits in the range of print parameters, and parameters from NN-GA are conducted, and the results show that the dimensional error is much lower using the printing parameters from NN-GA, and also show that the NN-GA is capable to promote the dimensional accuracy of 3DP and provide the reference for other forms AM technology.