Application of artificial neural network in predicting the performance of SPIF for aerospace grade AA7075-T6

Abstract

Single Point Incremental Forming (SPIF) is recently been used to prepare customized parts with complex geometries. The vertical CNC-milling machine and CNC code are generally used to carry out the forming operation. The material used in the present study is aerospace grade AA7075-T6 due to its inherent properties of high strength-to-weight ratio. The input parameters such as annealing temperature, step depth, tool shape, and the number of cut-out blanks were taken into consideration. The Output parameters like surface roughness and formability were recorded and evaluated by developing an Artificial Neural Network (ANN) based prediction model. Both single-output and double-output neural networks were established for this purpose. In case of one-output structure, the training function of trainlm, transfer function of logsig and 8 no. of neurons whereas, the training function of trainlm, transfer function of tansig and 6 no. of neurons are found to be suitable combination for surface roughness and formability respectively. On the other hand, in case of two-output structure, the training function of trainlm, transfer function of tansig and 10 no. of neurons are found to be suitable. The MAPE and R-values of the developed ANN model denote good agreement with the experimental results. The developed model will reduce the cost, effort, and time of young engineers and practitioners to select accurate parameters without performing expensive experimental runs.