Please use this identifier to cite or link to this item:
|Title:||High cycle fatigue life prediction of laser additive manufactured stainless steel : a machine learning approach||Authors:||Zhang, Meng
Goh, Phoi Chin
|Keywords:||Engineering::Mechanical engineering||Issue Date:||2019||Source:||Zhang, M., Sun, C.-N., Zhang, X., Goh, P. C., Wei, J., Hardacre, D., & Li, H. (2019). High cycle fatigue life prediction of laser additive manufactured stainless steel : a machine learning approach. International Journal of Fatigue, 128, 105194-. doi:10.1016/j.ijfatigue.2019.105194||Journal:||International Journal of Fatigue||Abstract:||Variations in the high cycle fatigue response of laser powder bed fusion materials can be caused by the choice of processing and post-processing strategies. The numerous influencing factors arising from the process demand an effective and unified approach to fatigue property assessment. This work examines the use of a neuro-fuzzy-based machine learning method for predicting the high cycle fatigue life of laser powder bed fusion stainless steel 316L. A dataset, consisting of fatigue life data for samples subjected to varying processing conditions (laser power, scan speed and layer thickness), post-processing treatments (annealing and hot isostatic pressing) and cyclic stresses, was constructed for simulating a complex nonlinear input-output environment. The associated fracture mechanisms, including the modes of crack initiation and deformation, were characterised. Two models, by employing the processing/post-processing parameters and the static tensile properties respectively as the inputs, were developed from the training data. Despite the diverse fatigue and fracture properties, the models demonstrated good prediction accuracy when checked against the test data, and the computationally-derived fuzzy rules agree well with understanding of the fracture mechanisms. Direct application of the model to literature results, however, yielded a range of prediction accuracies because of the variability in the reported data. Retraining the model by incorporating the literature results into the dataset led to improved modelling performance.||URI:||https://hdl.handle.net/10356/140927||ISSN:||0142-1123||DOI:||10.1016/j.ijfatigue.2019.105194||Rights:||© 2019 Elsevier. All rights reserved. This paper was published in International Journal of Fatigue and is made available with permission of Elsevier.||Fulltext Permission:||open||Fulltext Availability:||With Fulltext|
|Appears in Collections:||SC3DP Journal Articles|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.