Prediction and optimization of strain homogeneity in PTCAP process using ANFIS, ANN, and Taguchi approaches

Abstract

In this paper, the PTCAP (parallel tubular channel angular pressing) process as an SPD (severe plastic deformation) method is investigated. The effects of process parameters including channel angle, friction coefficient, and punch speed on strain homogeneity of the pure copper tube are studied. First, 9 experiments were designed by using Taguchi L9 orthogonal array. The experiments were carried out via the finite element method and the desired output was determined. The S/N (signal-to-noise) ratio and the ANOVA (analysis of variance) were used to determine the important parameters and calculate the contributions of each parameter. The results indicate that the channel angle has the most influence on strain homogeneity. In addition, a channel angle of 150 degrees, a friction coefficient of 0.05, and a punch speed of 9 mm/min led to the best result. Also, optimization results show almost a 6% improvement in strain homogeneity compared to conventional results. Finally, ANN (Artificial Neural Network) is used for predicting the output of the nine experiences. Afterwards, an ANFIS (Adaptive Neuro-Fuzzy Inference System) is also used for this prediction. Results show that the root square mean error in predicting the output using ANN is 5.51 × 10−3, and it reduced magnificently to 7.39 × 10−7 using the benefits of fuzzy in ANFIS.