An improved neural network model for prediction of mechanical properties of magnesium alloys

Abstract

An improved neural network model was developed for prediction of mechanical properties in the design and development of new types of magnesium alloys by refining the types of input variables and using a more reasonable algorithm. The results showed that the improved model apparently decreased the prediction errors, and raised the accuracy of the prediction results. Better preprocessing parameters were found to be [0.15, 0.90] for the tensile strength, [0.1, 0.9] for the yield strength, and [0.15, 0.90] for the elongation. When the above parameters were used, the relativity for predicition of strength was bigger than 0.95. By using improved ANN analysis, more reasonable process parameters and composition could be obtained in some magnesium alloys without addition of strontoum. magnesium alloys, neural network model, composition, mechanical properties BackgroundBecause of their low density, magnesium alloys have attracted an increasing interest in transportation, aeronautical and aerospace industries for the past decade, but their mechanical properties and processing performances still could not meet the needs of some important parts in vehicles and other important application fields [1][2][3][4] . Many new types of magnesium alloys are being developed in the world in order to further improve the mechanical properties and processing performances of the magnesium alloys [5][6][7] . However, conventional methods for developing new alloys need a lot of experimental work and take long time.Prediction of mechanical properties of engineering alloys is important for scientists and engineers, which can save not only cost but also time. However, due to the complex interconnections among chemical compositions and materials properties, conventional mathematical models are sometimes very complex to handle by the numerical techniques. In recent years, neural network models have been widely used in different metallurgical operations. Efforts have been made to use this technique for predicting the hot extrusion processes of AZ31 and AZ61 magnesium alloys and for investigating the influence of Y and Zn additions on the mechanical properties of Mg-Zn-Zr-Y alloys [8][9][10][11][12] . However, significant prediction errors in some cases call for further improvement of the neural network model for more accurate predictions. In the present work, an improved neural network model was developed for prediction of mechanical properties in the design and development of new types of magnesium alloys by refining the types of input variables and using a more reasonable algorithm.

Prediction of mechanical strength of magnesium alloy AZ31 with calcium addition using a neural network based model

“…The number of layers in the hidden layer can be defined along with the number of neurons. Though the accuracy of the model could not be predicted with the standard set of number of neurons and hidden layers [14,18,19], the weights, biases, number of neurons in the hidden layers and minimized errors are the factors which also influences the results. A proper training of input data with the corresponding output data yielding regression values more than 0.93 and least mean squared errors would be commonly accepted factors to predict best solution for the input of the particular network model.…”

“…As the mechanical properties, generally depend upon the composition of materials, grain size, textures and orientations, presence and absence of phases of metal oxides etc., it is very difficult to predict the properties before the specimen is prepared and tested experimentally. But the neural network training algorithm Levenberg-Marquardt showed remarkable results with minimum mean squared error and the data are tested and the results are validated for the performance of the algorithm [11][12][13][14][15][16].…”

In vitro and in vivo cytocompatibility evaluation of biodegradable magnesium-based stents: a review

“…No foreign material would be left behind, which could effectively prevent chronic inflammation, delayed allergic reactions and further invasive treatments at the same site [11,17]. The major limitation in the application of Mgbased stents is the rapid degradation associated with a decrease of mechanical performance [18,19]. At present, there are two methods to reduce the degradation rate and improve their mechanical properties.…”

Focuses of material science development in recent years

“…However, significant technological challenges remain, and magnesium is not a major structural material for many application. In recent years, many new alloys and innovative applications have been developed around the world [23][24][25][26][27][28][29][30][31][32]. Recent research and development status of wrought magnesium alloys in China is reviewed by Pan et al [23], and more attentions are paid to structure controlling, plastic processing, welding, surface treatment and product application.…”