Part Build Orientation Optimization and Neural Network-Based Geometry Compensation for Additive Manufacturing Process

Abstract

Significant advancements in the field of additive manufacturing (AM) have increased the popularity of AM in mainstream industries. The dimensional accuracy and surface finish of parts manufactured using AM depend on the AM process and the accompanying process parameters. Part build orientation is one of the most critical process parameters, since it has a direct impact on the part quality measurement metrics such as cusp error, manufacturability concerns for geometric features such as thin regions and small fusible openings, and support structure parameters. In conjunction with the build orientation, the cyclic heating and cooling of the material involved in the AM processes lead to nonuniform deformations throughout the part. These factors cumulatively affect the design conformity, surface finish, and the postprocessing requirements of the manufactured parts. In this paper, a two-step part build orientation optimization and thermal compensation methodology is presented to minimize the geometric inaccuracies resulting in the part during the AM process. In the first step, a weighted optimization model is used to determine the optimal build orientation for a part with respect to the aforementioned part quality and manufacturability metrics. In the second step, a novel artificial neural network (ANN)-based geometric compensation methodology is used on the part in its optimal orientation to make appropriate geometric modifications to counteract the thermal effects resulting from the AM process. The effectiveness of this compensation is assessed on an example part using a new point cloud to part conformity metric and shows significant improvements in the manufactured part's geometric accuracy.