About this Abstract |
Meeting |
2020 TMS Annual Meeting & Exhibition
|
Symposium
|
Aluminum Alloys, Processing and Characterization
|
Presentation Title |
Descriptors and Predictors: New Tools for the Predictive Modelling of Production Paths and the Properties of Aluminum-based End-products |
Author(s) |
Varuzan M. Kevorkijan, Irena Paulin, Crtomir Donik |
On-Site Speaker (Planned) |
Crtomir Donik |
Abstract Scope |
The processing-path vectors that describe industrial processes usually consist of a few hundreds of components. Therefore, modelling with such vectors is difficult and often lacks sufficient accuracy for industrial applications.
A possible solution is to replace them with the much simpler mathematical structures called descriptors. In the vector space used in this work, the descriptor is the distance between the vector of the selected and the vector of the average processing path.
The next step in finding a sufficiently strong correlation between the processing paths and the set of properties, is to calculate the conditional probabilities of all the available pairs. The predictor, the tool used in this study, is a triplet of data giving the probability higher than that required, that the processing path will result in a particular set of properties.
The algorithm was successfully validated in an industrial production of forging rods made from AA6082.
|
Proceedings Inclusion? |
Planned: Light Metals |