Journal
INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY
Volume 122, Issue 3-4, Pages 1143-1166Publisher
SPRINGER LONDON LTD
DOI: 10.1007/s00170-022-09916-4
Keywords
Machine learning; Additive manufacturing; Parameter optimization; Defect detection; Process-structure-property relationship
Ask authors/readers for more resources
Machine learning has become a mainstream idea in additive manufacturing, improving efficiency and product quality while creating autonomous frameworks. This review explores the application of machine learning in various areas of additive manufacturing, from material selection to process parameter optimization, and discusses its role in establishing process-structure-property relationships and defect detection.
Machine learning (ML) has undeniably turned into a mainstream idea by enhancing any system's throughput by allowing a more intelligent usage of materials and processes and managing their resultant properties. In industrial applications, usage of ML not only decreases the lead time of the manufacturing process involved but because of iterative steps of process parameters optimization, it also increases the quality and properties of the parts produced. Furthermore, ML provides an opportunity for creating completely or partially autonomous frameworks. A subset of ML, i.e., deep learning (DL), has capabilities of interpreting data in a layered pattern with little or no requirement of the labeled data for training. On the other hand, additive manufacturing (AM) offers benefits in designing intricated 3D shapes and gaining well-defined control over processing parameters, which eventually control the quality of a final product. This review discusses the utilization of ML techniques in various areas of AM ranging from the selection of material and alloy development to AM process parameter optimization. ML data training also helps in establishing the relation between AM process-structure-property relationship and defect detection in the printed objects. Consecutive steps of the process, i.e., data gathering, population establishment, model selection, training, and application, have been discussed. Also, certain challenges associated with the long-term incorporation of ML in the AM have been identified and their probable solutions have been provided.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available