Russian scientists have developed a method for assessing the strength of alloys using a neural network. This was reported by the press service of the RNF.
Most metals have a crystalline structure—that is, the atoms in them are precisely aligned. However, by rapidly cooling the melt, lattice formation can be prevented and the metal can be made amorphous. Amorphous alloys can have a number of advantages, be lighter and stronger than their crystalline counterparts, and are therefore often used in mechanical engineering and the manufacture of sports and medical equipment. To characterize the strength of the material, the Young’s modulus, determined experimentally by compressing or stretching a sample of the alloy of interest, is used. However, in the case of amorphous alloys, scientists still do not understand what this value depends on.
Now experts from Kazan Federal University have developed a neural network that can describe the relationship between various physical and chemical properties and predict the value of the Young’s modulus. To train the algorithm, the authors used data on more than 300 different alloys, including aluminum, copper, iron, and other metals.
The neural network determined that Young’s modulus is mainly affected by two indicators: the yield strength of the material and the glass transition temperature. The first value reflects the physical load at which the alloy begins to deform, and the second indicates the temperature at which, after cooling, the liquid melt solidifies into the corresponding solid amorphous alloy. Thus, using only these two parameters, the neural network determined the Young’s modulus for various compounds with up to 98% accuracy compared to the experimentally obtained values.
At the same time, it turned out that the chemical properties of the alloy, such as the amount and molecular weight of its constituent elements, do not affect the resistance to tension and compression. This is shown by the algorithm calculating the Young’s modulus based on these properties with an error of about 50%.
The authors hope their program will simplify the development of new metals for industry.
Previously, scientists found that female spiders fake dead so male partners aren’t afraid of being beaten.