Multidirectional forging (MULTIDIRE

Multidirectional forging (MULTIDIRECTIONAL FORGING) is the one of the severe plastic deformation process where in which a large plastic strain is induced in to the materials to get ultra-fine grain structure. It is more useful technique and simple in operation to produce ultra-fine grained large bulk samples [16–20]. Biomedical devices developed using bulk materials, the surface is very important because it forms an interface with biological envi ronment and is often important for integrity and mechanical success. In this aspect, altering a biomaterial surface is one of the techniques to increase corrosion resistance; it looks to be a very encouraging method to handle the above mentioned issue in Mg–Zn alloys. Ball burnishing is a novel surface modification technique carried out on material surface to enhance surface qualities and to induce compressive residual stresses which has attracted a substantial consideration in research [21–24]. This process comprises of a spherical ball that is pressed and rolled along the surface of the workpiece and economically affordable process because it can be performed on a CNC machine tool. Denkena et al. [25] reported that ball burnishing can control the corrosion rate and to attain desired degradation profile of Mg implants which is intended for various medical applications. Pu et al. [26] performed ball burnishing on AZ31B Mg alloy which produces fine grain structure and basal textured surface leading to improved corrosion performance. But there are many parameters are involved in the burnishing pro cess such as depth of press, speed, feed, burnishing force, ball material, ball size, number of pass and type of lubricant. Performing or varying all the parameters during experiment is very expensive and time consuming. To overcome these limitations, artificial neural network (ANN) (non-linear statistical model) is a reliable prediction technique that can be successfully employed in prediction of materials properties. The deep neural network refers to the type of ANN where in the system uses numerous node layers to derive high-level operations from inputs. Deep neural networks are computer programs that are biologically motivated to replicate human intelligence processes [27–29]. DNN model, which normally contains less hidden layers for relatively less data applications, has also attracted increased attentions in the manufacturing applications. By understanding data trends and associa tions, DNNs gain knowledge and learn (or have been trained) from experience rather than programming. It can self-regulate and fit various nonlinearities in the data series through training and learning, which provides a method with high quality and efficiency to calculate the optimal conditions for manufacturing processes [30–31]. This method is mainly used when the relationship among the studied variables is complex or when the knowledge of the physical correlations is limited. The interest in DNN modeling in the fields of materials science and physical metallurgy has increased rapidly [32–34]. Meimei Liu et al. [35] applied ANN for prediction and analysis of high velocity oxy fuel sprayed coating. Author reported that R-value of 0.99965 with a maximum error of 2.675% is achieved to verify the prediction perfor mance of the ANN model and the reliability and accuracy of the ANN model have been further verified by the test sets. Durmus et al. [36] investigated use of neural networks for the prediction of wear loss and surface roughness of AA6351 aluminum alloy. Author reported that the results obtained in ANN application are close to test results. Therefore by using trained ANN values, the intermediate results that are not obtained in the tests can be calculated.