In this paper an Artificial Intelligent approach that performs materials' tests and evaluates their properties based on Neural Networks is presented. Wide ranges from rigid and non-rigid to very limp materials are used as test samples for the compression and extension tests. A Robot as well as a universal testing machine are used for extension (for non-rigid materials) and compression (for rigid materials only) tests under variable temperatures and strain rate conditions. Destructive and non-destructive approaches are followed. The evaluation of the measured forces and the prediction of the materials' properties are based on Artificial Neural Networks that have been previously trained according the experts’ knowledge. The materials properties are expressed as quantitative and qualitative data. The results obtained showing that the neural network approach can accurately describe the material flow stress under the considered processing conditions as well as can classify the non-rigid materials according to their extensibility.
Evaluating and Calculating the Properties of Rigid and Non-rigid Materials Via Neural Networks / P. N., Koustoumpardis; N. A., Aspragathos; D'Addona, DORIANA MARILENA; Teti, Roberto. - STAMPA. - 6:(2008), pp. 217-222.
Evaluating and Calculating the Properties of Rigid and Non-rigid Materials Via Neural Networks
D'ADDONA, DORIANA MARILENA;TETI, ROBERTO
2008
Abstract
In this paper an Artificial Intelligent approach that performs materials' tests and evaluates their properties based on Neural Networks is presented. Wide ranges from rigid and non-rigid to very limp materials are used as test samples for the compression and extension tests. A Robot as well as a universal testing machine are used for extension (for non-rigid materials) and compression (for rigid materials only) tests under variable temperatures and strain rate conditions. Destructive and non-destructive approaches are followed. The evaluation of the measured forces and the prediction of the materials' properties are based on Artificial Neural Networks that have been previously trained according the experts’ knowledge. The materials properties are expressed as quantitative and qualitative data. The results obtained showing that the neural network approach can accurately describe the material flow stress under the considered processing conditions as well as can classify the non-rigid materials according to their extensibility.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.