|About this Abstract
||2021 TMS Annual Meeting & Exhibition
||Additive Manufacturing of Metals: Applications of Solidification Fundamentals
||Development of a True Porosity Distribution Evaluation and Pores’ Shape Assessment Method for Wire Arc Additive Menufactured Parts Using Neural Network
||Dmitrii Kurushkin, Igor Mushnikov, Ivan Kladov, Arthur Khismatullin, Irina Molodtsova, Oleg Panchenko
|On-Site Speaker (Planned)
In order to study the crystallization processes of the deposited metal and the formation of the manufactured structure, produced by wire arc additive manufacturing, it is necessary to solve the problem of porosity evaluation. One of the most modern methods for porosity evaluation is X-ray tomography. To provide a reconstruction of data, obtained with the use of X-ray tomography, there is a difficulty to separate intersecting clusters of pores, as well as to determine their true shape. Due to the high noise level of the original data, it is necessary to use specialized algorithms to solve the problem of separating intersecting pores. In this work, an algorithm based on a neural network was developed to obtain a complete picture of the distribution of pores from the data acquired using X-ray tomography.