Application of data mining for the identification of parts for additive manufacturing (original) (raw)

2020 International Conference on Intelligent Engineering and Management (ICIEM), 2020

Abstract

The objective of this study is to implement an AI-based method to identify parts in (ERP, EDM or) CAD systems that are especially attractive for Additive Manufacturing (AM). For this purpose, the potentials and limitations of AM are analyzed and formalized. A set of twelve indicators is developed that can be used to assess the complexity of a part as significant parameter for the use of AM. To evaluate these abstract indicators expressing part-complexity, a novel AI-supported procedure is developed. Two classifiers are deduced using a decision tree and an artificial neural network to evaluate and rank the indicators according to their performance. The resulting classifiers reveal that for both AI-methods a part identification on an accuracy level above 90 percent is possible by assessing seven indicators. Both classifiers show prediction accuracies in the same order of magnitude. The resulting AI tool is applied to a practical use case from plant engineering to demonstrate the industrial impact. Several parts of the plant are identified as potential AM parts. The evaluation of manufacturing cost of the most promising part reveal that AM can be beneficial compared to conventional manufacturing.

Georg Jacobs hasn't uploaded this paper.

Let Georg know you want this paper to be uploaded.

Ask for this paper to be uploaded.