Please use this identifier to cite or link to this item:
|Title:||Prediction performance of various numerical model training algorithms in solidification process of A356 matrix composites|
|Authors:||Shabani, Mohsen Ostad|
|Keywords:||Finite element;Artificial intelligence;Aluminum;Metal matrix composite|
|Abstract:||This paper reports the microstructural and mechanical properties of casting Al matrix composite such as porosity, hardness and tensile strength. The numerical model and finite element method are applied to simulate the solidification of the composites. The finite element analysis involves a number of steps such as finite-element discretization, imposition of boundary conditions and solution of assembled equations. The mathematical formulation of this solidification problem is given. The neural network predictions are directly compared with the experimentally obtained data to evaluate the learning performance. In this investigation the MAPE is used to evaluate the performance of model. The results show that Levenberg-Marquardt learning algorithm give the best prediction for UTS, hardness and porosity of A356 composite reinforced with B4C particulates.|
|ISSN:||0975-1017 (Online); 0971-4588 (Print)|
|Appears in Collections:||IJEMS Vol.19(2) [April 2012]|
Items in NOPR are protected by copyright, with all rights reserved, unless otherwise indicated.