Fatigue database of additively manufactured alloys

Abstract

Fatigue is a process of mechanical degradation that is usually assessed based on empirical rules and experimental data obtained from standardized tests. Fatigue data of engineering materials are commonly reported in S-N (the stress-life relation), ε-N (the strain-life relation), and da/dN-ΔK (the relation between the fatigue crack growth rate and the stress intensity factor range) data. Fatigue and static mechanical properties of additively manufactured (AM) alloys, as well as the types of materials, parameters of AM, processing, and testing are collected from thousands of scientific articles till the end of 2022 using natural language processing, machine learning, and computer vision techniques. The results show that the performance of AM alloys could reach that of conventional alloys although data dispersion and system deviation are present. The database (FatigueData-AM2022) is formatted in compact structures, hosted in an open repository, and analyzed to show their patterns and statistics. The quality of data collected from the literature is measured by defining rating scores for datasets reported in individual studies and through the fill rates of data entries across all the datasets. The database also serves as a high-quality training set for data processing using machine learning models. The procedures of data extraction and analysis are outlined and the tools are publicly released. A unified language of fatigue data is suggested to regulate data reporting for the fatigue performance of materials to facilitate data sharing and the development of open science.

Fatigue database of complex metallic alloys

“…A brief description of the workflow is given below. Only the key points and improvements of our methodology are reported here, and details can be found in our previous work on the fatigue data of AM alloys 43 . The performance of automated data extraction is assessed by the metrics, precision, recall, and F1 score, that are where TP denotes the true positive or the number of correctly-extracted data, FP is the false positive or the number of incorrectly-extracted data, and FN is the false negative or the number of data that are not extracted.…”

“…2a ), with an F1 score of 90% (Table 2 ). The fatigue data ( S - N , ε - N , and d a /d N -Δ K ) presented in scatter plots are extracted by an in-house MATLAB code IMageEXtractor (IMEX) 43 . Tables in XML/HTML files are parsed by table extractor 65 whereas those embedded in the PDFs are processed manually.…”

“…Experimentally measured 36 , 38 , 39 and theoretically predicted 40 , 41 data reported in the literature and research reports are valuable sources to understand and improve the performance of alloys. Text mining, image data processing, and artificial intelligence 37 , 42 techniques are expected to offer insights into the CPMP relationship as demonstrated in our recent work on AM alloys 43 . However, constructing such a large database involves processes of data generation, collection, publication, and maintenance, which are time-, cost-, and labor-intensive and beyond the capability of a single researcher or group.…”

Prediction method of longitudinal surface settlement caused by double shield tunnelling based on deep learning

“…Studies pertaining to longitudinal surface settlement utilise settlement information per ring, thereby significantly increasing the database capacity compared with studies pertaining to the maximum surface settlement 26 . However, the high excavation speed of the shield tunnelling method causes inadequate monitoring of surface settlement, resulting in a limited database capacity compared to other machine learning problems 32 , 33 . Ye et al 34 obtained surface settlement data by tunnelling under ancient towers and compared the prediction accuracy of databases with different capacities.…”

Prediction Method of Longitudinal Surface Settlement Caused by Double Shield Tunnelling Based on Deep Learning

“…Studies pertaining to longitudinal surface settlement utilise settlement information per ring, thereby significantly increasing the database capacity compared with studies pertaining to the maximum surface settlement [26] . However, the high excavation speed of the shield tunnelling method causes inadequate monitoring of surface settlement, resulting in a limited database capacity compared to other machine learning problems [31,32] . Ye, et al [33] obtained surface settlement data by tunnelling under ancient towers and compared the prediction accuracy of databases with different capacities.…”