Model Selection and Evaluation for Machine Learning: Deep Learning in Materials Processing
- PDF / 1,798,468 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 59 Downloads / 204 Views
TECHNICAL ARTICLE
Model Selection and Evaluation for Machine Learning: Deep Learning in Materials Processing Adam Kopper1 · Rasika Karkare2 · Randy C. Paffenroth3 · Diran Apelian4 Received: 15 July 2020 / Accepted: 28 August 2020 © The Minerals, Metals & Materials Society 2020
Abstract Materials processing is a critical subset of manufacturing which is benefitting by implementing machine learning to create knowledge from the data mined/collected and gain a deeper understanding of manufacturing processes. In this study, we focus on aluminum high-pressure die-casting (HPDC) process, which constitutes over 60% of all cast Al components. Routinely collected process data over a year’s time of serial production are used to make predictions on mechanical properties of castings, specifically, the ultimate tensile strength (UTS). Random Forest, Support Vector Machine (SVM), and XGBoost regression algorithms were selected from the machine learning spectrum along with a Neural Network, a deep learning method. These methods were evaluated and assessed and were compared to predictions based on historical data. Machine learning, including Neural Network, regression models do improve the predictability of UTS above that of predicting the mean from prior tests. Choosing the correct models to use for the data requires an understanding of the bias-variance trade-off such that a balance is struck between the complexity of the algorithms chosen and the size of the dataset in question. These concepts are reviewed and discussed in context of HPDC. Keywords Machine learning · Deep learning · Random forest · Support vector machine · Neural network · High pressure die casting · Principal component analysis · Bias-variance trade-off
Introduction A recent boom in machine learning has been sparked by continuous decrease in the cost of computer memory and increases in computing power [1, 2]. This, coupled with increased access to machine learning algorithms and open source software, has broadened the scope of interested parties beyond the early adopters like social media, banking, and marketing and retail sectors into manufacturing operations. Materials processing is a critical subset of manufacturing which is benefitting by implementing machine learning
* Adam Kopper [email protected] 1
Mercury Marine, Fond du Lac, WI 54935, USA
2
Data Science, WPI, Worcester, MA 01609, USA
3
Mathematical Sciences, Computer Science, Data Science, WPI, Worcester, MA 01609, USA
4
Materials Science and Engineering, UCI, Irvine, CA 92967, USA
to create knowledge from the data mined/collected and to gain a deeper understanding of manufacturing processes. Many materials manufacturing processes tend to be large scale in terms of production tonnage and units per hour. Efficiency is a core metric for materials processing plants. In thermally controlled processes, interruptions have significant downtime implications in returning the process to the operating temperature. In such an environment, sampling each unit of the product for the p
Data Loading...
